Abstract
Objective
Preoperative type and cross is often routinely ordered before elective endovascular aneurysm repair (EVAR), but the cost of this practice is high, and transfusion is rare. We therefore aimed to stratify patients by their risk of transfusion to identify a cohort in whom a type and screen would be sufficient.
Methods
We queried the targeted vascular module of the National Surgical Quality Improvement Program for all elective EVARs from 2011–2015. We included only infrarenal aneurysms and excluded ruptured aneurysms, and patients transfused within 72 hours preoperatively. Two-thirds of the cases were randomly assigned to a model derivation cohort, and one third to a validation cohort. We created and subsequently validated a risk model for transfusion within the first 24 hours of surgery (including intraoperatively), using logistic regression.
Results
Between 2011–2015, 4,875 patients underwent elective infrarenal EVAR, of which only 221 (4.5%) received a transfusion within 24 hours of surgery. The frequency of transfusion over the study period declined monotonously from 6.5% in 2011 to 3.2% in 2015. The factors independently associated with transfusion were preoperative hematocrit < 36% (Odds Ratio (OR) 3.4 [95% Confidence Interval 2.1 – 5.4], P < .001), aortic diameter (per centimeter increase: OR 1.2 [1.03 – 1.4], P = .02), preoperative dependent functional status (OR 2.5 [1.1–5.5], P = .03), and chronic obstructive pulmonary disease (OR 1.7 [1.04 – 2.9], P = .04). A risk prediction model based on these criteria produced a C-statistic of 0.69 in the prediction cohort and 0.76 in the validation cohort, and a Hosmer-Lemeshow Goodness of fit of 0.62 and 0.14 respectively. A score of < 3 out of 9, corresponding to a < 5% probability of transfusion, would avoid preoperative type and cross in 86% of patients. Of the 4,203 patients (86%) with a hematocrit > 36%, only 6 (0.1%) had a risk score of > 3.
Conclusion
Perioperative transfusion for EVAR is becoming increasingly uncommon and is predicted well by a transfusion risk score, or simply a hematocrit of < 36%. Application of this risk score would avoid unnecessary type and cross in the majority of patients, leading to significant savings in both time and cost.
Introduction
Since its introduction in 1991, endovascular aneurysm repair (EVAR) rapidly supplanted traditional open surgery as the predominant modality for abdominal aortic aneurysms (AAA).1–4 By 2008, EVAR accounted for 77% of the repairs performed in Medicare patients, and that number continues to rise.4 As EVAR proliferated, surgical techniques expanded to include percutaneous access, and the grafts and sheaths used improved dramatically in efficacy and profile. However, many elements of perioperative care lagged behind.
Compared to open aortic surgery, EVAR is associated with dramatically lower blood loss and transfusions.5,6 Prior reports document transfusion rates of less than 18% after EVAR, and most below 10%.5,7,8 This represents a marked improvement from open repair, which is associated with over 50% transfusion rates in some reports.5,8 These series are limited in that they use older data sets, include EVAR performed for ruptures, and mostly examine repairs via femoral cutdown.5,7,8 Since most of the blood loss during EVAR occurs in obtaining access or exchanging sheaths, these transfusion rates may not reflect contemporary practice.
Many blood banks still perform a type and cross before all EVARs, and many ensure that two or more units be in the operating room.9,10 The Maximum Surgical Blood Ordering Schedule (MSBOS) at our own institution calls for four units to be type and crossed for every EVAR. At Johns Hopkins, even after the implementation of an updated, sophisticated system to analyze their blood utilization, their MSBOS still calls for a type and cross for two units for every EVAR.9,11 The cost of this practice is significant. At our institution, a type and screen costs $325, and a type and cross adds an additional $126–$175, plus the costs of each additional unit of blood. Nationally, Medicare reimbursement is $136.36 for a simple type and screen, with an additional $209.49 if any antibodies are positive. Medicare reimburses an additional $104.37 for a type and cross, plus $99.65 for each unit of blood.12 Over 40,000 patients undergo AAA repair in the United States each year, the majority of which will be performed endovascularly, so any change in practice would lead to a substantial cost savings.
To our knowledge, there are few studies to date that examine the contemporary transfusion rates after elective EVAR, or validate the need for preoperative type and cross. In the only study to directly assess this topic, Mann and colleagues examined 203 elective EVARs from 2001–2010, and found a transfusion rate of 6%.10 However, this was a small sample of patients from 7 to 16 years ago at a relatively low volume, European center, and thus may not be generalizable to modern practice. Consequently, we decided to study modern transfusion practices after EVAR using a prospectively collected, national registry, and derive a risk prediction model to identify a low risk cohort of patients in whom a type and screen would be sufficient.
Methods
Data Source
We performed a retrospective cohort study using the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) and its targeted vascular module. As of 2015, there were 603 participating sites, 89 of which contributed to the targeted vascular module. At these sites, trained, clinical reviewers prospectively collect demographics, comorbidities, preoperative laboratory data, intraoperative details, and 30-day surgical outcomes according to a strict protocol. The data are routinely audited for accuracy and reliability. More information can be found in the NSQIP user guide.13 The data within NSQIP are retrospective and de-identified, and as such, the Beth Israel Deaconess Medical Center Institutional Review Board waived the need for informed consent.
Patients
We identified all patients undergoing intact, elective, infrarenal EVAR between 2011 and 2015 within the NSQIP database. We excluded patients with symptomatic or ruptured aneurysms, as well as patients transfused within 72 hours preoperatively, in order to isolate the transfusion requirements attributable to the procedure itself. Patients who underwent placement of iliac conduits were excluded. We randomly assigned two thirds of the cohort to a model derivation cohort, and the remaining one third to a validation cohort.
Outcomes
Our primary outcome was transfusion intraoperatively and within 24 hours postoperatively. NSQIP does not capture the number of units transfused or distinguish between patients transfused within 24 hours postoperatively and those transfused intraoperatively.
Statistical Analysis
We randomly assigned two-thirds of the cohort to a model derivation cohort, and the remainder to a validation cohort. We created a logistic regression model for transfusion within 24 hours, and included factors which were determined a priori based on their plausible associations with need for transfusion. Our initial model included age; preoperative hematocrit, platelets, and INR; body mass index (BMI); open cutdown versus percutaneous femoral access; aortic diameter; diabetes; dialysis; chronic kidney disease; chronic obstructive pulmonary disease; congestive heart failure; non-independent functional status; hypertension; smoking; white race; female sex; degree of preoperative dyspnea; steroid use; bleeding disorders; and the year of the operation. As our goal was a parsimonious model, we selected those factors that were independently associated with transfusion within 24 hours with a threshold of P < .05 to include in our final model. This parsimonious risk prediction model was compared to the original model using the likelihood ratio test. We then assessed our model performance using the C-statistic of the receiver operating curve, and the Hosmer-Lemeshow Goodness-of-fit test.
The point scoring system in our risk model was developed using the βeta coefficients in our logistic regression model using the method described in the Framingham Heart Study.14 The predictive ability of this model was tested in the derivation dataset and subsequently in the validation dataset using the C-statistic of the receiver operating curve, and the Hosmer-Lemeshow Goodness-of-fit test. All statistical analyses were performed using Stata 14.2 (StataCorp, College Station, Tex).
Results
Baseline Demographics
We identified 4,875 elective, infrarenal EVARs performed between 2011 and 2015. Patient characteristics are presented in Table I. Patients who were transfused within 24 hours were more likely to be female and older, have lower BMI, and have more comorbid conditions.
Table I.
Baseline Characteristics. (s.d.: standard deviation, BMI: body mass index, IQR: Interquartile Range, ASA: American Society of Anesthesiologists, COPD: chronic obstructive pulmonary disease)
|
Not transfused on POD0 (n=4.654) |
Transfused on POD0 (n=221) |
P value | |
|---|---|---|---|
| Characteristic | n(%) or median [IQR] | ||
| Female | 777(17) | 60 (27) | <.01 |
| Age | 74 (8.5) | 77 (8.6) | <.01 |
| White | 3856 (94) | 192 (92) | 0.3 |
| BMI | 28 [25–32] | 26 [23–30] | <.01 |
| BMI Category: | <.01 | ||
| underweight (<18.5) | 48 (1) | 8 (4) | <.01 |
| normal (18.5>=BMI <25) | 1124 (24) | 79 (36) | |
| overweight (25>=BMI <30) | 1879 (40) | 78 (35) | |
| Obese (30>=BMI >40) | 1402 (30) | 47 (21) | |
| morbidly obese (>=40) | 201 (4) | 9 (4) | |
| ASA 1 | 4 (<1) | 0 | <.01 |
| ASA 2 | 275 (6) | 6 (3) | |
| ASA 3 | 3282 (71) | 131 (59) | |
| ASA 4 | 1081 (23) | 84 (38) | |
| ASA 5 | 3 (<1) | 0 | |
| Smoking | 1408 (30) | 53 (24) | 0.051 |
| Hypertension | 3699 (79) | 190(86) | 0.02 |
| Independent | 4557 (98) | 199 (90) | <.01 |
| CHF | 53 (1) | 4 (2) | 0.33 |
| COPD | 815(18) | 60 (27) | <.01 |
| CKD (4–5) | 125 (3) | 18 (8) | <.01 |
| Hemodialysis | 31 (1) | 5 (2) | 0.02 |
| Diabetes | 750(16) | 38 (17) | 0.64 |
| Bleeding Disorder | 508 (11) | 42(19) | <.01 |
| Dyspnea | 0.17 | ||
| At Rest | 54 (1) | 4 (2) | |
| Steroids | 197 (4) | 13 (6) | 0.23 |
| Moderate Exertion | 773(17) | 45 (20) | |
| No | 3827 (82) | 172 (78) | |
| Aneurysm Specific | 0.02 | ||
| Small | 2163 (46) | 84 (38) | |
| >55mm | 2491 (54) | 137 (62) | |
| Preoperative: | |||
| Hematocrit: | 41 [38–44] | 36 [32–41] | <.01 |
| < 30 | 75 (2) | 30 (14) | |
| 30–36 | 491 (11) | 75 (34) | |
| >36 | 4,087 (88) | 116 (52) | |
| Albumin | 4 [3.7–4.2] | 3.9 [3.5–4.2] | <.01 |
| INR>1.5 | 130 (4) | 11 (6) | 0.17 |
| Platelets | 198 [165–241] | 198 [154–248] | 0.07 |
| Operative: | |||
| Access: | <.01 | ||
| Perc conv open | 35 (1) | 14 (6) | |
| Bilateral cutdown | 2521 (54) | 138 (62) | |
| Unilateral cutdown | 374 (8) | 20 (9) | |
| Percutaneous | 1716 (37) | 49 (22) | |
| Conv to open | 12 (<1) | 8 (4) | <.01 |
| Hypogastric embolization | 243 (5) | 20 (9) | 0.02 |
| Hypogastric revascularization | 105 (2) | 12 (5) | <.01 |
Transfusion Rates
During the study period, only 221 patients (4.5%) received a transfusion either intraoperatively or within 24 hours, with 301 patients (6.2%) ultimately receiving a transfusion within 30 days. The frequency of perioperative transfusion declined monotonously over the study period from 6.5% in 2011 to 3.2% in 2015. The factors independently associated with transfusion were preoperative hematocrit < 36% (Odds Ratio (OR) 3.4 [95% Confidence Interval 2.1 – 5.4], P < .001), aortic diameter (per centimeter increase: OR 1.2 [1.03 – 1.4], P = .02), preoperative dependent functional status (OR 2.5 [1.1 – 5.5], P = .03), and chronic obstructive pulmonary disease (OR 1.7 [1.04 – 2.9], P = .04). These were the factors selected for our parsimonious model.
Risk Model
Our parsimonious model was not statistically significantly different from the full model by likelihood ratio test (P = 0.40). The model resulted in a C-statistic of 0.73 and a goodness-of-fit of 0.84. The C-statistic is a measure of the discrimination of a model, while the goodness-of-fit assesses the calibration. The βeta coefficients were used to assign point values to each of the variables involved to create a risk scoring model (Table II). Points ranged from 0 to 5, with a potential risk score of 0 to 9. The risk score had a C-statistic of 0.69 and a goodness-of-fit of 0.62 in the model derivation cohort. We then applied the risk score to the validation cohort, and found a C-statistic of 0.76 and a goodness-of-fit of 0.14. The risk scores, their individual and cumulative frequencies, and the associated risk of transfusion are presented in Table III. Notably, a score of < 3 out of 9 was associated with a less than 5% risk of transfusion. If we applied this rather generous cutoff as the threshold for type and cross rather than type and screen, we would avoid type and cross in 86% of patients.
Table II.
Risk Score Works Cited
| Criteria | Points |
|---|---|
|
| |
| Hematocrit | |
| <30 | 5 |
| 30–36 | 3 |
| >36 | 0 |
| Diameter | |
| <5.5 | 0 |
| >5.5 | 1 |
| COPD | |
| Yes | 1 |
| No | 0 |
| Independent Functional Status? | |
| Yes | 0 |
| No | 2 |
Table III.
Transfusion risk, Frequency, and Cumulative Frequency by Risk Score
| Total Points | Predicted Risk of Transfusion | Freaqency | Cumulative Freaqency |
|---|---|---|---|
| 0 | 1.97% | 41.54% | 41.54% |
| 1 | 3.02% | 37.37% | 78.91% |
| 2 | 4.62% | 7.24% | 86.15% |
| 3 | 6.99% | 4.86% | 91.02% |
| 4 | 10.45% | 5.11% | 96.12% |
| 5 | 15.34% | 1.87% | 97.99% |
| 6 | 21.96% | 1.37% | 99.36% |
| 7 | 30.40% | 0.39% | 99.75% |
| 8 | 40.41% | 0.12% | 99.88% |
| 9 | 51.29% | 0.12% | 100% |
Hematocrit Alone
The most significant predictor of postoperative transfusion was preoperative hematocrit. As such, we elected to study whether or not a preoperative hematocrit alone was sufficient to predict transfusion. We applied a threshold of 36% based on previous work as well as the inflection point in our data where risk of transfusion appeared to increase.15–18 In our cohort, there were 4,203 patients (86%) who had a hematocrit > 36%. Only 6 of those patients (0.1%) had a risk score of > 3 (which would qualify them for type and cross by our model), and only 116 (2.7%) were transfused. In contrast, 105 (16%) of the 672 patients with hematocrit ≤ 36 underwent transfusion. This single predictor had a negative predictive value of 97%.
Cost Implications
The Medicare reimbursement for a type and cross with two units is $440.02, while a type and screen is only $136.36 if no antibodies are present.12 Thus, eliminating the type and cross from an EVAR would represent a savings of $303.66 per case. With approximately 40,000 EVARs in the US per year, using our risk score to remove the crossmatch from 86% represents a potential savings of $10,445,904 annually.19 This conservative estimate represents only the direct product acquisition cost, and does not account for the entirety of the transfusion and blood-bank related expenditures such as leukocyte reduction, irradiation, blood bank staff, and the opportunity costs and overhead costs of delivering two crossmatched units to the operating room for each case. Prior studies demonstrated that these direct costs account for only 21–32% of the total expenses related to transfusions, so the cost savings are likely dramatically higher.20,21
Discussion
Our study used a large, national, clinically validated registry to examine the transfusion rates after EVAR in contemporary practice, and develop a risk score to identify those patients at higher risk. Notably, transfusion during and after EVAR is much lower than previous series, and our data demonstrate that this risk continues to decline over time. Indeed, only 3% of patients who underwent EVAR in 2015 received an intraoperative transfusion or were transfused within 24 hours. This highlights the changes in practice patterns that were not captured adequately in previous works. Percutaneous access is now commonplace, sheaths and devices are lower profile and better designed with valves to reduce blood loss, and more restrictive transfusion protocols were implemented as more data emerged about the risks of blood transfusions.
Interestingly, the type of access (percutaneous versus open femoral cutdown) was not associated with postoperative transfusion. This is likely due to both the low overall need for transfusion, as well as the blood loss during sheath exchange or closure device failure offsetting the decrease in blood loss with percutaneous access. As we are only examining a four-year period, it is beyond the scope of our data to extrapolate a learning curve for percutaneous access, but this is certainly an interesting future direction of study. We suspect that percutaneous access will be associated with less transfusions in the future as surgeons become more familiar with percutaneous technique, and sheaths and closure devices continue to improve.
In addition to the improvements from a surgical perspective, modern blood banks have made substantial strides as well. Electronic type and cross adopted by many centers allows for rapid, emergent crossmatches in a matter of minutes. With the exception of patients with antibodies (< 2% in most studies), these computerized systems can use a type and screen to identify crossmatched blood for hemorrhaging patients in minutes.22–25 Massive transfusion protocols initially designed for trauma patients have improved hemorrhage management in other cases as well.26–29 These factors have made it easier to rapidly obtain blood in cases where only a type and screen is available, further obviating the need for routine type and cross.
With this changing environment in mind, we aimed to develop a model designed to identify a low risk cohort of patients in whom a type and cross was unnecessary. Our model accurately stratifies patients based on their risk of perioperative transfusion, and would allow for significant cost savings if implemented. We chose a threshold of 5% based on prior work related to avoiding type and screen in patients undergoing carotid endarterectomy.30 This threshold is likely overly conservative, and can be modified depending on the characteristics of the individual centers. For example, centers with robust, well-developed massive transfusion protocols and electronic type and cross can apply higher thresholds, whereas centers without those capabilities can use a lower score.
Our goal was to create a simple score that is easy to remember and implement at the point of care. With this in mind, we created a maximally parsimonious model with only four variables that did not deviate significantly from the full model in terms of predictive ability. Preoperative hematocrit accounted for over half the possible points in the score, so we examined the predictive utility of hematocrit alone. Indeed, a hematocrit threshold of 36% both avoided type and cross in the majority of patients, and only missed a small fraction of patients who were transfused.
These data must be interpreted in the context of the retrospective nature of the study design, as well as the limitations of the NSQIP database. We lack data on preoperative or postoperative medications such as antiplatelet agents or anticoagulants. In addition, NSQIP does not record the amount of blood transfused, merely whether or not one or more transfusions occurred. We also do not know why the transfusions occurred, so there is no way of distinguishing between patients transfused for certain hemoglobin or hematocrit thresholds, and those who underwent massive transfusion because of life-threatening hemorrhage. However, we were not as concerned with the reason for a transfusion as much as whether one or more occurred. Our goal was to create a model that utilized preoperative risk factors to predict intraoperative or postoperative transfusion of any kind or indication. These were elective, uncomplicated, infrarenal EVARs only. Ruptured, symptomatic or complex EVARs including snorkels or fenestrations, as well as juxtarenal or suprarenal aneurysms were excluded, and thus our score cannot be applied to these cases.
It is important to note that we do not advocate our risk score as a method of eliminating type and screen for EVAR, merely type and cross. Although cost saving is an important factor in our modern healthcare system, the potential outcomes are asymmetric. We chose a very conservative threshold of < 5% risk of transfusion, because the potential for life-threatening hemorrhage without readily available blood cannot be easily weighed in a cost-effectiveness analysis. Consequently, centers without rapid, massive transfusion protocols and electronic crossmatching should take these considerations into account when deciding on their individual cutoffs.
Conclusion
Perioperative transfusion for EVAR is becoming increasingly uncommon and is predicted well by a transfusion risk score, or simply a hematocrit of < 36%. Application of this risk score would avoid unnecessary type and cross in the majority of patients, leading to significant savings in both time and cost.
Type of Research
Retrospective analysis of the prospectively collected data of the targeted vascular module NSQIP registry.
Take Home Message
Only 4.5% of 4,875 patients who underwent endovascular aortic aneurysm repair (EVAR) needed blood transfusion within 24 hours. Factors associated with transfusion included preoperative hematocrit, dependent functional status, aortic diameter and COPD. A risk prediction model identified patients with < 5% risk of transfusion and would avoid preoperative type and cross in 86% of patients, leading to cost savings.
Recommendation
The authors propose a risk scoring system that would avoid type and cross in over 86% of patients who undergo EVAR, and lead to substantial cost savings.
Acknowledgments
Supported by the Harvard-Longwood Research Training in Vascular Surgery NIH T32 Grant 5T32HL007734-22
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Works Cited
- 1.Parodi JC, Palmaz JC, Barone HD. Transfemoral Intraluminal Graft Implantation for Abdominal Aortic Aneurysms. Ann Vasc Surg [Internet] 1991 Nov;5(6):491–9. doi: 10.1007/BF02015271. [cited 2016 Nov 13]; Available from: http://linkinghub.elsevier.com/retrieve/pii/S0890509606615410. [DOI] [PubMed] [Google Scholar]
- 2.Giles KA, O’Malley AJ, Cotterill P, Pomposelli FB, Landon BE, Schermerhorn ML. RR15. Age Related Trends in Abdominal Aortic Aneurysms: Repairs, Rupture, and Death Rates in the United States Medicare Population. J Vasc Surg [Internet] 2009 May;49(5):S50. [cited 2016 Sep 12] Available from: http://linkinghub.elsevier.com/retrieve/pii/S0741521409004431. [Google Scholar]
- 3.Schermerhorn ML, O’Malley AJ, Jhaveri A, Cotterill P, Pomposelli F, Landon BE. Endovascular vs. Open Repair of Abdominal Aortic Aneurysms in the Medicare Population. N Engl J Med [Internet] 2008 Jan 31;358(5):464–74. doi: 10.1056/NEJMoa0707348. [cited 2016 Sep 12] Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa0707348. [DOI] [PubMed] [Google Scholar]
- 4.Schermerhorn ML, Buck DB, O’Malley AJ, Curran T, McCallum JC, Darling J, et al. Long-Term Outcomes of Abdominal Aortic Aneurysm in the Medicare Population. N Engl J Med [Internet] 2015 Jul 23;373(4):328–38. doi: 10.1056/NEJMoa1405778. [cited 2016 Sep 12] Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa1405778. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Raval MV, Eskandari MK. Outcomes of elective abdominal aortic aneurysm repair among the elderly: Endovascular versus open repair. Surgery [Internet] 2012 Feb;151(2):245–60. doi: 10.1016/j.surg.2010.10.022. [cited 2017 Feb 17] Available from: http://www.ncbi.nlm.nih.gov/pubmed/21244863. [DOI] [PubMed] [Google Scholar]
- 6.Davenport DL, O’Keeffe SD, Minion DJ, Sorial EE, Endean ED, Xenos ES. Thirty-day NSQIP database outcomes of open versus endoluminal repair of ruptured abdominal aortic aneurysms. J Vasc Surg [Internet] 2010 Feb;51(2):305–309.e1. doi: 10.1016/j.jvs.2009.08.086. [cited 2017 Feb 17] Available from: http://www.ncbi.nlm.nih.gov/pubmed/19939609. [DOI] [PubMed] [Google Scholar]
- 7.Turnbull IC, Criado FJ, Sanchez L, Sadek M, Malik R, Ellozy SH, et al. Five-year results for the Talent enhanced Low Profile System abdominal stent graft pivotal trial including early and long-term safety and efficacy. J Vasc Surg [Internet] 2010 Mar;51(3):537–544.e2. doi: 10.1016/j.jvs.2009.09.039. [cited 2017 Feb 17] Available from: http://www.ncbi.nlm.nih.gov/pubmed/20206803. [DOI] [PubMed] [Google Scholar]
- 8.Kauvar DS, Martin ED, Givens MD. Thirty-Day Outcomes after Elective Percutaneous or Open Endovascular Repair of Abdominal Aortic Aneurysms. Ann Vasc Surg [Internet] 2016 Feb;31:46–51. doi: 10.1016/j.avsg.2015.10.009. [cited 2017 Feb 17] Available from: http://www.ncbi.nlm.nih.gov/pubmed/26658092. [DOI] [PubMed] [Google Scholar]
- 9.Frank SM, Rothschild JA, Masear CG, Rivers RJ, Merritt WT, Savage WJ, et al. Optimizing preoperative blood ordering with data acquired from an anesthesia information management system. Anesthesiology [Internet] 2013 Jun;118(6):1286–97. doi: 10.1097/ALN.0b013e3182923da0. [cited 2016 Sep 20] Available from: http://www.ncbi.nlm.nih.gov/pubmed/23695091. [DOI] [PubMed] [Google Scholar]
- 10.Mann K, Sim I, Ali T, Chong P, Leopold P, Hatrick A, et al. Removing the need for crossmatched blood in elective EVAR. Eur J Vasc Endovasc Surg [Internet] 2012 Mar;43(3):282–5. doi: 10.1016/j.ejvs.2011.11.026. [cited 2016 Oct 7] Available from: http://www.ncbi.nlm.nih.gov/pubmed/22182351. [DOI] [PubMed] [Google Scholar]
- 11.Frank SM, Oleyar MJ, Ness PM, Tobian AAR. Reducing unnecessary preoperative blood orders and costs by implementing an updated institution-specific maximum surgical blood order schedule and a remote electronic blood release system. Anesthesiology [Internet] 2014 Sep;121(3):501–9. doi: 10.1097/ALN.0000000000000338. [cited 2016 Sep 20] Available from: http://www.ncbi.nlm.nih.gov/pubmed/24932853. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Centers for Medicare and Medicaid Services. 2016 Proposed Medicare Blood Product Payment [Internet] 2016 [cited 2017 Apr 4]. Available from: https://www.aabb.org/advocacy/reimbursementinitiatives/Documents/2016-HOPPS-Proposed-Rule-Summary.pdf.
- 13.ACS National Surgical Quality Improvement Program [Internet] [cited 2017 Feb 17]. Available from: https://www.facs.org/quality-programs/acs-nsqip.
- 14.Sullivan LM, Massaro JM, D’Agostino RB. Presentation of multivariate data for clinical use: The Framingham Study risk score functions. Stat Med [Internet] 2004 May 30;23(10):1631–60. doi: 10.1002/sim.1742. [cited 2017 Feb 17] Available from: http://www.ncbi.nlm.nih.gov/pubmed/15122742. [DOI] [PubMed] [Google Scholar]
- 15.Wu W-C, Rathore SS, Wang Y, Radford MJ, Krumholz HM. Blood Transfusion in Elderly Patients with Acute Myocardial Infarction. N Engl J Med [Internet] 2001 Oct 25;345(17):1230–6. doi: 10.1056/NEJMoa010615. [cited 2017 Feb 13] Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa010615. [DOI] [PubMed] [Google Scholar]
- 16.Ma JZ, Ebben J, Xia H, Collins AJ. Hematocrit Level and Associated Mortality in Hemodialysis Patients. doi: 10.1681/ASN.V103610. [DOI] [PubMed] [Google Scholar]
- 17.Sabatine MS, Morrow DA, Giugliano RP, Burton PBJ, Murphy SA, McCabe CH, et al. Association of Hemoglobin Levels With Clinical Outcomes in Acute Coronary Syndromes. Circulation. 2005;111(16) doi: 10.1161/01.CIR.0000162477.70955.5F. [DOI] [PubMed] [Google Scholar]
- 18.Sarhane KA, Flores JM, Cooney CM, Abreu FM, Lacayo M, Baltodano PA, et al. Preoperative Anemia and Postoperative Outcomes in Immediate Breast Reconstructive Surgery: A Critical Analysis of 10,958 Patients from the ACS-NSQIP Database. Plast Reconstr surgery Glob open [Internet] 2013 Aug;1(5):e30. doi: 10.1097/GOX.0b013e3182a18c6f. [cited 2017 Feb 13] Available from: http://www.ncbi.nlm.nih.gov/pubmed/25289224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Karthikesalingam A, Vidal-Diez A, Holt PJ, Loftus IM, Schermerhorn ML, Soden PA, et al. Thresholds for Abdominal Aortic Aneurysm Repair in England and the United States. N Engl J Med [Internet] 2016 Nov 24;375(21):2051–9. doi: 10.1056/NEJMoa1600931. [cited 2017 Feb 17] Available from: http://www.nejm.org/doi/10.1056/NEJMoa1600931. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Shander A, Hofmann A, Ozawa S, Theusinger OM, Gombotz H, Spahn DR. Activity-based costs of blood transfusions in surgical patients at four hospitals. Transfusion [Internet] 2010 Apr;50(4):753–65. doi: 10.1111/j.1537-2995.2009.02518.x. [cited 2017 Mar 12] Available from: http://doi.wiley.com/10.1111/j.1537-2995.2009.02518.x. [DOI] [PubMed] [Google Scholar]
- 21.Shander A, Hofmann A, Gombotz H, Theusinger OM, Spahn DR. Estimating the cost of blood: past, present, and future directions. Best Pract Res Clin Anaesthesiol. 2007;21(2):271–89. doi: 10.1016/j.bpa.2007.01.002. [DOI] [PubMed] [Google Scholar]
- 22.Smith HM, Shirey RS, Thoman SK, Jackson JB. Prevalence of clinically significant red blood cell alloantibodies in pregnant women at a large tertiary-care facility. Immunohematology [Internet] 2013;29(4):127–30. [cited 2017 Feb 18] Available from: http://www.ncbi.nlm.nih.gov/pubmed/24689681. [PubMed] [Google Scholar]
- 23.Zaman S, Chaurasia R, Chatterjee K, Thapliyal RM. Prevalence and Specificity of RBC Alloantibodies in Indian Patients Attending a Tertiary Care Hospital. Adv Hematol [Internet] 2014;2014:749218. doi: 10.1155/2014/749218. [cited 2017 Feb 18] Available from: http://www.ncbi.nlm.nih.gov/pubmed/25386192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Makroo RN, Bhatia A, Hegde V, Chowdhry M, Thakur UK, Rosamma NL. Antibody screening & identification in the general patient population at a tertiary care hospital in New Delhi, India. Indian J Med Res [Internet] 2014 Sep;140(3):401–5. [cited 2017 Feb 18] Available from: http://www.ncbi.nlm.nih.gov/pubmed/25366208. [PMC free article] [PubMed] [Google Scholar]
- 25.Tormey CA, Fisk J, Stack G. Red blood cell alloantibody frequency, specificity, and properties in a population of male military veterans. Transfusion [Internet] 2008 Oct;48(10):2069–76. doi: 10.1111/j.1537-2995.2008.01815.x. [cited 2017 Feb 18] Available from: http://www.ncbi.nlm.nih.gov/pubmed/18631165. [DOI] [PubMed] [Google Scholar]
- 26.Stanworth SJ, Davenport R, Curry N, Seeney F, Eaglestone S, Edwards A, et al. Mortality from trauma haemorrhage and opportunities for improvement in transfusion practice. Br J Surg [Internet] 2016 Mar;103(4):357–65. doi: 10.1002/bjs.10052. [cited 2017 Feb 18] Available from: http://www.ncbi.nlm.nih.gov/pubmed/26841720. [DOI] [PubMed] [Google Scholar]
- 27.Etchill E, Sperry J, Zuckerbraun B, Alarcon L, Brown J, Schuster K, et al. The confusion continues: results from an American Association for the Surgery of Trauma survey on massive transfusion practices among United States trauma centers. Transfusion [Internet] 2016 Oct;56(10):2478–86. doi: 10.1111/trf.13755. [cited 2017 Feb 18] Available from: http://www.ncbi.nlm.nih.gov/pubmed/27515056. [DOI] [PubMed] [Google Scholar]
- 28.Schuster KM, Davis KA, Lui FY, Maerz LL, Kaplan LJ. The status of massive transfusion protocols in United States trauma centers: massive transfusion or massive confusion? Transfusion [Internet] 2010 Feb 11;50(7):1545–51. doi: 10.1111/j.1537-2995.2010.02587.x. [cited 2017 Feb 18] Available from: http://www.ncbi.nlm.nih.gov/pubmed/20158684. [DOI] [PubMed] [Google Scholar]
- 29.Treml AB, Gorlin JB, Dutton RP, Scavone BM. Massive Transfusion Protocols. Anesth Analg [Internet] 2017 Jan;124(1):277–81. doi: 10.1213/ANE.0000000000001610. [cited 2017 Feb 18] Available from: http://www.ncbi.nlm.nih.gov/pubmed/27749352. [DOI] [PubMed] [Google Scholar]
- 30.Stangenberg L, Curran T, Shuja F, Rosenberg R, Mahmood F, Schermerhorn ML. Development of a risk prediction model for transfusion in carotid endarterectomy and demonstration of cost-saving potential by avoidance of “type and screen”. J Vasc Surg [Internet] 2016 Jul 15; doi: 10.1016/j.jvs.2016.04.059. [cited 2016 Sep 20]; Available from: http://www.ncbi.nlm.nih.gov/pubmed/27432200. [DOI] [PMC free article] [PubMed]