Methods: A retrospective analysis using hospital administrative data examined patients (=18 years) who underwent orthopaedic surgery and received Evicel, Vitagel, or no haemostat between 1/1/2009 and 11/30/2009. A 1:1 (Evicel:Vitagel) and 1:3 (Evicel:no haemostat) match was conducted on surgery type, teaching hospital status, and propensity scores.

Results: There were 316 patients identified for the Evicel versus Vitagel analysis and 1,808 patients for the Evicel versus no haemostat analysis. Patients in the Vitagel cohort were 6.8 times more likely (OR=6.8, 95% CI 1.6–28.7) and patients in the no haemostat cohort were 1.6 times more likely (OR=1.6, 95% CI 1.08–2.5) to have a hospital readmission than the Evicel cohort. Average total index hospital cost was lower for Evicel than for Vitagel ($16,704 versus $18,192, p<0.001). Evicel had similar total costs compared to no haemostat use ($17,387 versus $17,389, p=0.18).

Conclusions: When used during inpatient orthopaedic surgery, Evicel was associated with lower readmission rates compared with Vitagel or no haemostat use. Total costs were significantly lower for Evicel than Vitagel but similar to no haemostat.

J.Orthopaedics 2011;9(1)e5

Key words:orthopaedics, fibrin sealant, fibrin glue, readmission, hospital cost

Introduction:

Patients undergoing knee or hip arthroplasty are at risk of significant blood loss1, which mainly occurs during the postoperative period2 and may necessitate blood transfusions that can impact the duration and cost of initial hospitalization3. However, the extent to which inadequate haemostasis contributes to hospital readmission rates and the overall cost of inpatient care in orthopaedic surgery settings have not been characterized.

Fibrin sealants are a class of topical haemostatic agents that provide adequate haemostasis by locally increasing fibrinogen and thrombin levels, thereby increasing the rate of blood clotting4. Various products are commercially available, which may contain fibrinogen, thrombin, factor XIII, and calcium in some form and combination5. Two common products in the fibrant sealant class used in orthopaedic surgeries are Evicel™ Fibrin Sealant (Human) (Johnson & Johnson Wound Management, a division of Ethicon, Inc., Somerville, NJ, USA) and Vitagel™ Fibrin Sealant (Orthovita, Inc., Malvern, PA, USA). Both fibrin sealants are indicated for use in surgical procedures as an adjunct to haemostasis when control of bleeding by standard surgical procedures is ineffective or impractical. Evicel is available as an all-human formulation, while Vitagel is used in conjunction with the patient’s own plasma but contains bovine thrombin6,7. Payors and hospital administrators are becoming increasingly concerned about costs and readmission rates among hospitalized patients, which are also important components of hospital quality measures8. Although fibrin sealants are effective in controlling local and diffuse bleeding, the perceived high cost of this type of surgical haemostat has become a concern that could limit their use4. Therefore, it is important to understand how the use of different fibrin sealants impact surgical outcomes, including readmission rates and overall hospital costs. The purpose of this study was to examine these outcomes following inpatient orthopaedic surgeries for patients receiving Evicel versus Vitagel or no haemostat use.

Methods

Data Source

This retrospective matched cohort study was conducted utilizing hospital administrative data from the Premier Perspective Database™ (Premier, Inc., Charlotte, NC, USA). This database contains inpatient, clinical, drug utilization, and hospital billing data from more than 600 hospitals throughout the United States. Premier collects data from participating hospitals in its healthcare alliance. The Premier healthcare alliance was formed for hospitals to share knowledge, improve patient safety, and reduce risks. Participation in the Premier healthcare alliance is voluntary. Although the database excludes federally funded hospitals (eg, Veterans Affairs), the hospitals included are nationally representative based on bed size, geographic location, and teaching hospital status. The database contains a date-stamped log of all billed items by cost-accounting department, including: medications; laboratory, diagnostic, and therapeutic services; and primary and secondary diagnoses for each patient. Identifier-linked enrollment files provide demographic and payor information. Data between July 1, 2008 and December 31, 2009 were utilized for this analysis.

Study Design and Sample Selection

Patients aged 18 years or older with an inpatient surgery of interest (ie, knee replacement [ICD-9-CM procedure codes of 81.54, 00.80–00.83] or hip replacement [ICD-9-CM procedure codes of 80.05, 81.51, 81.52, 81.53, 00.70–00.73, 00.85–00.87]) were eligible for study inclusion. The patient’s first chronologically occurring hospitalization of interest between January 1, 2009 and November 30, 2009 was selected as the index hospitalization. Patients were required to receive either Evicel or Vitagel on the day of surgery during the index hospitalization. A control group of patients who received no haemostat on the day of surgery was also selected. Patients were excluded if they had a prior hospitalization within 30 days of index hospitalization or were transferred from another hospital or unknown source. Eligible patients were then classified into cohorts based on therapy received.

Outcome Measures

The primary outcomes of interest included all-cause readmission rates and total cost of care. Readmission was defined as a subsequent hospitalization for any reason within 30 days after discharge. The average total cost of care (medical and pharmacy) incurred by the hospitals for patients in each cohort was calculated for the index hospitalization plus 30 days post-discharge. Since the study was conducted from an institutional perspective, costs (rather than charges) were used. Medical costs included room and board, professional fees, surgery supplies, and laboratory services (including radiology and pathology testings). Pharmacy costs included acquisition cost of the drug dispensed. A sub-analysis was conducted calculating the cost of care for only the index hospitalization.

Statistical Analysis

Baseline demographics and clinical characteristics were described with standard summary statistics (means and proportions). Covariates were compared using chi-square test for categorical variables and t-tests or Mann Whitney tests for continuous variables. Initial analysis revealed baseline differences across the cohorts. To control for possible bias, the Evicel cohort was matched 1:1 to the Vitagel cohort and 1:3 to the no haemostat cohort based on individual propensity scores, surgery type (hip versus knee replacement), and teaching hospital status. Propensity scores for each individual were obtained from a logistic regression model measuring the probability of receiving Evicel. Covariates included in the propensity score (measured at baseline or during the 6 months prior to index hospitalization) were gender, age, race, provider area, hospital location (ie, rural versus urban), payor type, Charlson comorbidity index9, all payor refined-diagnosis related group (APR-DRG) mortality risk score, APR-DRG severity level, time spent in the operating room, prior inpatient hospitalization, prior haemostat use, prior erythropoietin use, and selected comorbidities (specifically thrombocytopenia, diabetes, obesity, chronic obstructive pulmonary disease, hypertension, renal disease, congestive heart failure, cancer, myocardial infarction, non-myocardial infarction, coronary disease, deep vein thrombosis, and liver cirrhosis). After match, covariates were compared for each cohort versus the Evicel cohort by performing McNemar’s test for categorical variables and paired t-test or Wilcoxon test for continuous variables.

Crude readmission rates and cost measures for the matched cohorts were computed. Differences in readmission rates were evaluated using multivariate conditional logistic regression controlling for the presence of post-surgical blood transfusions and erythropoietin use and hospital length of stay along with covariates that were statistically different after matching. Generalized linear models using a gamma distribution with a log link were utilized to assess differences in cost of care controlling for covariates that were statistically different after matching. All statistical analyses tested a 2-sided hypothesis of no difference between treatment cohorts at a significance level of 0.05 and were carried out using SAS statistical software (version 9.1).

Results
Study Population

A total of 126,842 patients met the inclusion criteria, including 459 who received Evicel, 972 who received Vitagel, and 125,411 who were not treated with an adjunctive haemostat product. At baseline, patient and hospital characteristics differed significantly between cohorts (Table 1). Post-matching, 316 patients were identified for the Evicel (n = 158) versus Vitagel (n = 158) comparison and 1,808 patients for the Evicel (n = 452) versus no haemostat (n = 1,356) comparison. Patient and hospital characteristics for these matched cohorts are displayed in Table 2. Although matching reduced many cohort differences, certain patient and hospital characteristics were still significantly different. Specific differences included bed size when comparing either the Vitagel or no haemostat cohort to the Evicel cohort and Charlson comorbidity index, presence of cancer, time spent in the operating room, and APR-DRG mortality risk score between the no haemostat and Evicel cohorts.

Outcomes

Patients in the Vitagel cohort were 6.8 times more likely to be readmitted to the hospital within the first 30 days post-discharge compared to patients in the Evicel cohort (12.7% versus 3.8%; odds ratio [OR]=6.8, 95% confidence interval [CI] 1.6–28.7). Those in the no haemostat cohort versus Evicel cohort were 1.6 times more likely to require hospital readmission (10.5% versus 8.6%; OR=1.6, 95% CI 1.1–2.5). The total cost of care by cohort for the index hospitalization and index hospitalization plus 1 month post periods are summarized in Table 3. Mean total costs for index hospitalization were significantly lower for the Evicel cohort than for the Vitagel cohort ($16,704 versus $18,192; p<0.001), as were those for index hospitalization plus 1 month post period ($17,350 versus $19,752; p<0.001). The primary driver of the cost difference was medical costs. For the Evicel versus no haemostat comparison, the average total cost of care for both periods (index hospitalization: $17,387 versus $17,389 [p=0.16]; index hospitalization plus 1 month post: $18,847 versus $18,601 [p=0.19]) were similar. However, medical costs were significantly lower for Evicel for index hospitalization and for index hospitalization plus 30 days post-discharge compared to the no haemostat cohort.

Discussion
Insurance companies, accrediting organizations, and hospital administrators have become increasingly interested in using quality measures as a tool to improve patient care and ultimately reduce cost of care. One such measure that is becoming important is 30-day readmission rates. In fact, the Centers for Medicare & Medicaid Services, which collects these data for certain disease areas, considers 30-day readmission rates an “outcome of care” measure to show how well a hospital is doing in preventing complications, educating patients about their care needs, and helping patients make a smooth transition from the hospital to home or another type of care facility.

In this retrospective study, patients receiving Evicel had lower 30-day all-cause hospital readmission rates than patients receiving Vitagel or no adjunct haemostat therapy. Total hospitalization costs, based on both medical and pharmacy costs and calculated for the initial hospitalization as well as for the initial hospitalization and 30 days post-discharge, were significantly reduced for the Evicel cohort compared to the Vitagel cohort but not when comparing the Evicel and no adjunct haemostat cohorts.

Overall, there is a paucity of data regarding the impact of inadequate haemostasis on the duration of initial hospitalization, rate of readmission, and overall hospital costs following joint replacement surgery. In a US retrospective series of 23,514 patients undergoing total hip replacement, major knee surgery, or hip fracture replacement, for which reported outcomes were the duration and cost of hospitalization (with no information pertaining to readmission), the incidence of major bleeding (defined as fatal bleeding, nonfatal bleeding at a critical site, reoperation due to bleeding, or overt bleeding with bleeding index =2) was 2.6%3. Multivariate analysis revealed that major bleeding associated with a longer length of stay (6.1 days versus 4.3 days for patients without bleeding; adjusted difference of 1.8 days [95% CI 1.5–2.0 days]) as well as higher mean total inpatient charges ($25,669 versus $18,076; adjusted difference of $7,593 [95% CI $6,622–$8,646])3. To date, limited prospective randomized controlled clinical trials have evaluated the use of fibrin sealants specifically in total joint replacement settings2,10-12. In a small pilot study of 80 patients undergoing hip arthroplasty, Vivostat patient-derived fibrin sealant (Vivostat A/S, Alleroed, Denmark) was not shown to significantly reduce intraoperative blood loss, drain volumes, transfusion requirements, or duration of hospitalization versus no use of topical haemostatic treatment, with mean hospitalization of 7 days per group10. Conversely, other series have demonstrated the potential for fibrin sealants (various products) to reduce perioperative blood loss among patients undergoing total hip11,12 or knee2 replacement, although none has reported on the duration or cost of initial hospitalization or readmission. The infrequent use of Evicel or Vitagel in this study (with 1% of the population meeting the inclusion criteria having received one of these agents) may reflect, at least in part, uncertainties regarding the cost:benefit ratio in patients undergoing joint replacement surgery due to the limited data.

To our knowledge, this is the first published study designed to compare clinically relevant outcomes pertaining to hospitalization-related costs and readmission rates for surgical patients receiving all-human versus bovine thrombin-containing/patient-derived fibrin sealants versus no adjunctive haemostat. Since our study was conducted from an institutional perspective, costs (rather than charges) were used, representing the actual costs to treat the patient and including room and board, professional fees, surgery, supplies, pharmacy services, and laboratory services. Regarding medical and pharmacy costs, both were significantly lower with Evicel versus Vitagel during the index hospitalization period, whereas only medical costs were significantly reduced during the index hospitalization plus 30 days post-discharge period. The reduction in medical costs with Evicel relative to no adjunct haemostat use, despite lack of differences between these cohorts with respect to total hospitalization costs, is a noteworthy finding, suggesting that a reduction in bleeding-related consequences afforded by Evicel may have offset the cost of treatment in this analysis.

As with any research, our study has limitations that warrant discussion. First, this study analyzed administrative data using a retrospective, nonrandomized study design rather than prospectively within a comparative randomized clinical trial. Second, quality of the administrative data, as well as missing data or coding errors, could potentially affect the statistical differences seen in this analysis. Specifically, readmission rates and costs may be underestimated if patients sought care outside the Premier network of hospitals. However, we assume that the underestimation of readmission rates and costs would equally affect each cohort, as there is no systematic reason to assume otherwise. Third, differences in the initial cohorts at baseline may have contributed to observed differences in outcomes, despite attempts to address this potential bias by matching the cohorts on demographic and clinical characteristics, and then controlling for remaining covariates in the statistical analysis. Despite the limitations of the analysis, it is important to reiterate that the primary goal of our investigation was to observe readmission rates economic outcomes in real-world practice. This goal was accomplished using the Premier Perspective Database, while matching criteria and statistical models strengthened the validity of the comparison.

Conclusion:

ACKNOWLEDGMENTS

Funding for this study and resultant publication was provided by Ethicon Inc. Authors X. Ye, M.F.T. Rupnow, and J. Hammond are employees of Ethicon, Inc. Authors M. Shah and E. Farrelly are employed by Xcenda, a consultant to Ethicon.

References