Secondary Logo

Promoting a Restrictive Intraoperative Transfusion Strategy: The Influence of a Transfusion Guideline and a Novel Software Tool

Picton, Paul, MBChB, MRCP, FRCA; Starr, Jordan, MD; Kheterpal, Sachin, MD, MBA; Thompson, Aleda M. L., MS; Housey, Michelle, MPH; Sathishkumar, Subramanian, MD, FRCA; Dubovoy, Timur, MD; Kirkpatrick, Nathan, BA; Tremper, Kevin K., MD, PhD; Engoren, Milo, MD; Ramachandran, Satya Krishna, MD

doi: 10.1213/ANE.0000000000002704
Blood Management: Original Clinical Research Report
Free

BACKGROUND: The effect of neither transfusion guidelines nor decision support tools on intraoperative transfusion has been previously evaluated. The University of Michigan introduced a transfusion guideline in 2009, and in 2011, the Department of Anesthesiology developed a transfusion decision support tool. The primary aim of this study was to assess the associations of the transfusion guideline and the optional use of the software transfusion tool with intraoperative behaviors; pretransfusion hematocrit assessment (whether or not a hematocrit was checked before each red cell unit) and restrictive red cell use (withholding transfusion unless the hematocrit was ≤21%).

METHODS: This was a before–after retrospective study without a concurrent control group of patients transfused 1–3 units of red cells intraoperatively. Three phases were studied to provide data both before and after the implementation of the transfusion guideline and the intraoperative software tool. Within each phase, trends of checking hematocrits before transfusion and restrictive transfusion were charted against time. F tests were used to measure differences of slopes. The difference between means of each phase was measured using Mann-Whitney U tests. Independent associations were measured using mixed-effects multivariable logistic regression. A secondary outcome analysis was conducted for 30-day mortality, myocardial infarction, renal injury, and their combination.

RESULTS: The transfusion guideline was associated with increased pretransfusion hematocrit evaluation (67.4%, standard deviation [SD] 3.9 vs 76.5%, SD 2.7; P < .001) and restrictive transfusion practice (14.0%, SD 7.4 vs 33.3%, SD 4.4; P = .001). After adjustment for confounders, the guideline phase was independently associated with increased hematocrit checking (odds ratio, 1.72; 95% confidence interval, 1.46–2.03; P < .001) and restrictive red cell transfusion (odds ratio, 2.95; 95% confidence interval, 2.46–3.54; P < .001). The software tool was not associated with either transfusion behavior. There was no significant change in the rate of renal injury (16.06%), myocardial injury (4.93%), 30-day mortality (5.47%), or a composite (21.90%).

CONCLUSIONS: The introduction of a transfusion guideline was independently associated with increased intraoperative pretransfusion hematocrit assessment and restrictive transfusion. The use of a software tool did not further influence either behavior.

From the Department of Anesthesiology, University of Michigan Medical School, Ann Arbor, Michigan.

Published ahead of print December 15, 2017.

Accepted for publication October 20, 2017.

Funding: This work was supported by grants from the University of Michigan Department of Anesthesiology.

The authors declare no conflicts of interest.

Reprints will not be available from the authors.

Address correspondence to Paul Picton, MBChB, MRCP, FRCA, Department of Anesthesiology, University of Michigan Medical School, 1500 E Medical Center Dr, Ann Arbor, MI 48103. Address e-mail to ppicton@med.umich.edu.

KEY POINTS

  • Question: Are transfusion guidelines and/or software support tools associated with changes in transfusion-related behavior in the operating room?
  • Findings: The introduction of a transfusion guideline was independently associated with increased intraoperative pretransfusion hematocrit assessment and restrictive transfusion while the implementation of software tool did not further influence either behavior.
  • Meaning: Transfusion guidelines are associated with desirable transfusion behaviors within the operating room.

More than 13 million units of red cells are transfused each year within the United States at a mean average cost of $218/unit.1 Approximately 50% of these transfusion events occur within surgical specialties or critical care1–3; anesthesiologists are likely major determinants of the nationally transfused volume. Although severe anemia is associated with high perioperative mortality,4 red cell transfusion is also harmful,5 being independently associated with infection, lung injury, and death.5 The definition of restrictive transfusion during surgery is controversial given that active bleeding naturally requires a higher hemoglobin trigger and target. However, liberal transfusion strategies have not provided superior outcome for adults6 or children7 in critical care, for patients undergoing cardiac surgery on bypass,8 or for patients with risk factors for ischemic heart disease undergoing hip fracture repair.9 Liberal transfusion has also been found to increase 30-day mortality when compared to restrictive red cell transfusion for critically ill patients with Acute Physiology and Chronic Health Evaluation II scores <20,6 critically ill adults <55 years of age,6 and for patients presenting with acute gastrointestinal hemorrhage.10

The combinations of questionable efficacy, adverse risk profile, resource availability, and cost have driven the development of transfusion guidelines throughout the world. Within the University of Michigan Health System, a transfusion guideline was released in March 2009. As a departmental quality improvement initiative, we additionally developed a tool within our electronic anesthesia system (Centricity; General Electric Healthcare, Waukesha, WI) to facilitate communication and evidence-based transfusion practice. Clinical decision support has been shown to improve blood utilization within hospital systems,11 and previously published studies have evaluated the impact of education and decision support on red cell transfusion in the perioperative setting12; education appeared more effective than decision support. However, no study to date has investigated the effect of transfusion practice improvement interventions specifically within the unique environment of the operating room.

The primary aim of this study was to assess the associations of the hospital-wide transfusion guideline and optional use of the software transfusion tool on compliance with: (1) checking a hematocrit before transfusion (whether or not a hematocrit was checked before each red cell unit); and (2) adopting restrictive transfusion practice (withholding transfusion unless the hematocrit was ≤21%) within the operating room. The potential adverse effect of changes in transfusion behavior was assessed by evaluating changes in the rates of 30-day mortality, myocardial injury, and renal injury.

Back to Top | Article Outline

METHODS/DESIGN

This was a before–after retrospective study without a concurrent control group approved by the institutional review board of the University of Michigan, Ann Arbor, the presentation of which adheres to the applicable Enhancing the QUAlity and Transparency Of health Research (EQUATOR) guidelines. Patient-specific informed consent was not required. Patients <18 years of age and those undergoing cardiac surgery were excluded. To focus on the use of discretionary red blood cell (RBC) administration, only patients receiving between 1 and 3 intraoperative RBC units were included.

The intraoperative electronic records (Centricity) of adult patients undergoing a surgical procedure between April 1, 2005, and July 1, 2014, were interrogated. The date ranges were selected to provide data for patient groups both before and after the implementation of the hospital transfusion guideline and the development of an intraoperative software tool. Three phases were therefore studied: the baseline phase, the guideline phase, and the tool phase. For the baseline phase, the period from the start of the study, April 1, 2005, to the introduction of the hospital transfusion guideline, March 24, 2009, was assessed. For the guideline phase, the period after the introduction of the hospital transfusion guideline, March 24, 2009, but before the introduction of transfusion tool, March 31, 2011, was evaluated. For the tool phase, data from the period after the introduction of transfusion tool, March 31, 2011, to the end of the study period, July 1, 2014, were included.

The launch of the University of Michigan Transfusion Guideline was accompanied by hospital-wide education. The central and relevant recommendations for the purpose of this study are as follows: (1) to avoid red cell transfusion in patients with hemodynamically stable anemia unless the hemoglobin concentration is <7 g/dL; and (2) to transfuse red cells as single units whenever possible. Such an approach is supported by transfusion guidelines published by the AABB.13 During the guideline phase, discussion of estimated blood loss (EBL) became fully integrated into the preanesthesia verification and presurgical “time out.” The Department of Anesthesiology transfusion tool initiative became operational, March 31, 2011, after education and publicity: grand round lectures, announcements, e-mail communication, and in-room teaching. The goals of the software tool, embedded within our anesthesia information management system (AIMS), were to encourage compliance with the hospital guideline and to enhance communication between anesthesia and surgical teams surrounding transfusion decisions. Anesthesia providers are prompted to identify patients for whom the surgical team anticipated a blood loss of >500 mL or >10% circulating volume (Figure 1A) and to discuss the transfusion target with the faculty anesthesiologist and the surgical team. The target hematocrit range is selected from a pick list, and if a hematocrit >24% is chosen, a reason must be recorded from a second pick list (Figure 1B). Use of this process was not made mandatory.

Figure 1

Figure 1

Intraoperative data pertinent to transfusion were acquired: transfused red cell units, preoperative hemoglobin, hematocrit with the time of testing relative to the start of each transfused unit, final hematocrit, and for the post-tool group, intended target hematocrit and reasons for noncompliance with the transfusion guideline were evaluated. For the purpose of pretransfusion hematocrit checking, we arbitrarily considered it reasonable to base transfusion decisions on hematocrit values measured up to 1 hour before transfusion and also considered that hematocrits recorded in the electronic medical record up to 10 minutes after transfusion had commenced were likely available to the anesthesia team before transfusion. Laboratory values were derived from either the central laboratory or arterial and venous point of care testing. Providers were not alerted to the availability of centrally measured tests.

Back to Top | Article Outline

Study Outcomes

We defined transfusion as “compliant” with hematocrit testing before transfusion when results were documented within the 70-minute window for every unit transfused for a particular patient. The nadir hematocrit for transfused patients was evaluated as a proxy to transfusion trigger throughout the entire study period. The independent associations with pretransfusion hematocrit checking and a pretransfusion hematocrit ≤21% were evaluated throughout each phase. We assessed the safety of intraoperative transfusion practice change by evaluating 30-day mortality, perioperative myocardial injury, perioperative renal injury, and a collapsed composite of all 3 (yes for any of the 3) throughout the entire study period.

Back to Top | Article Outline

Statistical Analysis

Statistical analysis was performed utilizing SAS version 9.3 (SAS Institute, Cary, NC) and SPSS version 21.0 (IBM, Somers, NY). Basic descriptive statistics were calculated for demographics and relevant case data. Pearson χ2 or Fisher exact tests (for categorical variables) and Mann-Whitney U tests (for continuous variables) were used to assess baseline univariate clinical differences between groups for those compliant with pretransfusion hematocrit check and those with a pretransfusion hematocrit ≤21%. Normality of continuous measures was assessed using the Kolmogorov–Smirnov test.

To evaluate (1) compliance with pretransfusion hematocrit check and (2) transfusion with a pretransfusion hematocrit ≤21% over the study period, we charted trend against time and used an F test to measure the difference of the slope within each phase compared to zero as well as the difference in slopes between each phase. Time was considered in quarters within each phase. The difference in means between each phase was measured using Mann-Whitney U tests.

To further evaluate the association of the transfusion guideline and computerized tool with (1) compliance with pretransfusion % hematocrit check and (2) transfusion with a pretransfusion hematocrit ≤21%, we developed 2 mixed-effects multivariable logistic regression models. Clinical associations were analyzed to better understand factors affecting clinician-driven transfusion behavior.

Back to Top | Article Outline

Model Variables.

Fixed effects included were patient age, gender, surgical service, emergent case, American Society of Anesthesiologists status, history of coronary artery disease, body mass index (BMI), preoperative hemoglobin value, case duration (in hours), handover of care by any staff at any time during the case, handover of anesthesia attending at any time during the case, EBL for the entire case (in tertiles), number of units transfused, study phase, and the use of intraoperative transfusion tool. A random effect of the anesthesia attending was used to account for variability among providers.

Back to Top | Article Outline

Model Diagnostics.

Measures of effect size for fixed effects were reported as adjusted odds ratios (ORs) and 95% confidence intervals (CIs).14 Variables selected for model inclusion based on statistical significance and clinical relevance were evaluated for collinearity by analysis of pairwise Pearson correlation matrices. If the absolute value (positive or negative) of the correlation coefficient was <0.7, high correlation was excluded and variables were eligible for model inclusion. The overall predictive capability of both models was assessed using the bias-corrected C-statistics15 derived from bootstrapped Somer D estimates.16P values of .05 were considered statistically significant.

Back to Top | Article Outline

Missing Data.

No missing data were imputed. Patients with missing data were excluded from analysis. Descriptive statistics were computed to account for differences in missing data for each outcome; those patients whom were subject to hematocrit evaluation before each unit transfused versus those who were not and those patients transfused at hematocrit ≤21% versus those who were transfused at a higher hematocrit threshold.

Back to Top | Article Outline

Thirty-Day Mortality, Myocardial Injury, and Renal Injury.

To determine the relationship among 30-day postoperative mortality, myocardial injury, renal injury, and a collapsed composite of all 3 (yes to any of the 3) and time across the full study period, a secondary analysis was conducted. Myocardial injury was defined as postoperative troponin >0.3 ng/dL within 7 days of surgery, and renal injury was defined as an absolute increase in serum creatinine >0.3 mg/dL or an increase of >50% from baseline; baseline creatinine was taken as the most recent value measured within 30 days before surgery and was compared to the peak value measured within the first 72 hours postoperatively. Time was considered in quarters, and the trend across time was analyzed using F tests to measure the difference of the slope within each phase compared to zero as well as the difference in slopes between each phase. The difference in means between each phase was measured using Mann-Whitney U tests. Using a Bonferroni correction, a P value of .0125 was considered statistically significant for the 30-day mortality, myocardial injury, renal injury, and the composite variable analyses.

Back to Top | Article Outline

Sample Size and Power.

Power was calculated using the actual sample size. A sample size of 2984 in the baseline phase, 1253 in the guideline phase, and 1926 in the transfusion tool phase provides >90% power to detect a difference of 6% in both the proportion compliance with hematocrit checking and the proportion of restrictive transfusion rate (hematocrit ≤21) between successive phases assuming an α error of .05.

Back to Top | Article Outline

RESULTS

Study Populations and Missing Data

There were 10,042 patients who received 1–3 units of packed RBCs between April 1, 2005, and July 1, 2014. After exclusions (cardiac surgery: 2711, transfusion did not occur intraoperatively: 369, missing data: 772), 6163 patients were included for analysis; there were 2984 within the baseline phase, 1253 within the transfusion guideline phase, and 1926 within the transfusion tool phase. The transfusion tool was used for 620 (32%) patients within the transfusion tool phase. BMI (5.48%) and preoperative hematocrit (4.54%) were the variables most likely to be missing data, with no difference between those patients for whom a pretransfusion hematocrit was checked for every unit and those for whom it was not. For patients transfused at a pretransfusion hematocrit ≤21%, there was less missing BMI data (4.75% vs 7.81%; P < .001) and more missing preoperative hemoglobin data (5.12% vs 2.72%; P < .001) when compared to patients with a pretransfusion hematocrit of >21%.

Table 1

Table 1

The mean number of RBC units transfused was 1.77, standard deviation (SD) 0.69, during the baseline phase; 1.71, SD 0.70, during the guideline phase; and 1.67, SD 0.70, during the software tool phase. There was a statistically significant difference between the mean RBC units transfused between the baseline phase and the guideline phase (P = .006) but not between the guideline phase and the software tool phase (P = .104). Of the study population, all of whom received 1–3 units RBCs, fewer patients were transfused >1 unit RBCs in the guideline phase (56.5%) compared to the baseline phase (61.7%) P = .001 with an additional reduction in the software tool phase (53.1%) P = .044. Demographic information and univariate associations for patients transfused during the 3 study phases are presented in Table 1.

Back to Top | Article Outline

Checking Hematocrit Before Transfusion

We found that compliance for pretransfusion hematocrit check increased from the baseline period 67.4%, SD 3.9, to 76.5%, SD 2.7 (P < .001), in the guideline phase, but no further increase in compliance was noted after the introduction of the software tool 74.2%, SD 3.9 (P = .173). Within each phase, the compliance did not change (the slopes did not differ from zero, Figure 2A). Within each phase, the compliance was constant (the slopes did not differ from zero, Figure 2A).

Figure 2

Figure 2

After adjustment for confounders in a multivariable logistic regression model (Figure 2B), we found that the guideline phase was independently associated with an increased frequency of hematocrit checking before each red cell unit transfused compared to the baseline phase (OR, 1.72; 95% CI, 1.46–2.03; P < .001) and that the software tool phase was not associated with further increased frequency of hematocrit checking before transfusion when compared to the guideline phase (OR, 0.97; 95% CI, 0.80–1.17; P = .718). However, higher preoperative hemoglobin, greater EBL, case duration >2 hours, American Society of Anesthesiologists class ≥III, emergency surgery, and primary surgeon from transplant surgery were independently associated with higher compliance. Tests for collinearity revealed no highly correlated variables. Compliance did not vary by attending anesthesiologist (covariance parameter, 0.02; standard error, 0.03).

Back to Top | Article Outline

Transfusion With Hematocrit ≤21

Figure 3A represents the percentage of transfusions undertaken with a pretransfusion hematocrit ≤21% by quarter. The mean % patients transfused at a pretransfusion hematocrit ≤21% increased from 14.0%, SD 7.4, in the baseline phase to 33.3%, SD 4.4, in the guideline phase (P = .001) but did not increase further in the software tool phase (33.1%, SD 6.4; P = .97). During the 16 quarters of the baseline phase, the percentage of patients transfused with hematocrit ≤21% increased (slope = 1.3; P < .001). The percentage then further increased going to the guideline phase, but it did not increase further during the guideline or software tool phases (slopes not different from zero).

Figure 3

Figure 3

After adjusted multivariable analysis (Figure 3B), we found that the guideline phase was independently associated with increased restrictive intraoperative red cell transfusion when compared to the baseline phase (OR, 2.95; 95% CI, 2.46–3.54; P < .001) and that a further increase in restrictive transfusion was not evident within the software tool phase (OR, 1.02; 95% CI, 0.84–1.24; P = .812). Other independent associations with restrictive transfusion included transfused volume, case duration, emergency surgery, and the transplant surgery service. Transfusion was more likely to occur at % hematocrit threshold >21, rather than ≤21, for patients with coronary artery disease and for patients with a higher preoperative hemoglobin. Tests for collinearity revealed no highly correlated variables. The covariance parameter estimate for the random effect of the attending anesthesiologist was 0.10, standard error 0.04, implying that some anesthesiologists followed restrictive transfusion recommendations, others did not.

Back to Top | Article Outline

Selected Transfusion Targets and Reasons for Transfusion to a Hematocrit >24

Table 2

Table 2

Table 3

Table 3

After the introduction of the AIMS transfusion tool, most anesthesia providers selected a transfusion target hematocrit of 21%–24% (Table 2). For those selecting a transfusion target hematocrit of >24% (160 of 620 patients), the most common reasons selected from the pick list were coronary artery disease and surgeon preference (Table 3).

Back to Top | Article Outline

Thirty-Day Mortality, Myocardial Injury, and Renal Injury

Overall, within this transfused population, the 30-day mortality was 5.47%, myocardial injury occurred in 4.93%, and renal injury occurred in 16.06% with a composite of 21.90%. There was no significant change in the rate of renal injury, myocardial injury, or 30-day mortality during the entire study period. Similarly, there was no change in the occurrence of the composite outcomes between and during the 3 phases as measured by the means and the trends.

Back to Top | Article Outline

DISCUSSION

We studied the effect of the introduction of a transfusion guideline and the subsequent deployment of a software tool on: (1) compliance with checking pretransfusion hematocrit; and (2) restrictive transfusion. We found that the release of a hospital-wide transfusion guideline was independently associated with improved intraoperative compliance with pretransfusion hematocrit assessment and restrictive transfusion with adjusted ORs of 1.72 and 2.95, respectively. However, the addition of the software tool embedded within an AIMS system was not independently associated with further improvements in either pretransfusion hematocrit checking or restrictive transfusion. While the incidence of restrictive transfusion increased within patients transfused 1–3 units of packed RBCs during surgery from 14.0% in the baseline phase to 33.3% in the guideline phase, this was not associated with changes in composite morbidity and mortality. This study cannot determine a cause and effect relationship but, importantly, the greater usage of restrictive transfusion within the operating room was not associated with potentially ischemic outcomes from lower arterial oxygen content. We found that even before the guidelines were implemented, there was a trend of increased restrictive transfusion practices. This may be related to prerelease awareness of and education about the guidelines or awareness and response to the published literature supporting restrictive transfusion. With the implementation of the guideline, there was an abrupt further increase in compliance.

Software programs including real-time intraoperative automated alerts have been shown effective for the management of hyperglycemia,17,18 lung-protective ventilation strategies,19 improving the documentation of blood pressure,20 and have been used for intraoperative decision support.21 Use of the tool was not mandated, and the lack of an alert mechanism or real-time decision support22 may have limited its effects. It is possible that practitioners simply do not like to be guided by computers and probable that the prior implementation of the transfusion guideline and accompanying education limited the measured effect of the transfusion tool. This is similar to Zuckerberg et al12 who also found that education had a greater impact than computerized decision support. A more intrusive computer alert may have been more effective.11 It is unlikely, however, that 1 single system, guideline or event changes practice but rather multiple events23 move practitioners and organizations toward change. Indeed, the publication of landmark studies9,10,24 during the tool phase did little to change practice at our institution. The software tool may have been helpful to prevent decay of desirable transfusion-related behavior after the introduction of the transfusion guideline.

Restrictive transfusion was less likely to occur in patients with coronary artery disease. Although controversial, some studies have suggested that patients with cardiac disease do better with higher transfusion thresholds;25 such studies may influence transfusion decision making and may contribute to the apparent safety of restrictive transfusion in the “real-life” setting evaluated here. Doctors likely do not just follow guidelines but tailor treatment for individual patients. As shown by the covariance parameter estimates for the random effect of the attending anesthesiologist, we found little variability between attendings in their decisions to check a hematocrit but there was greater variability between attendings in their decisions to transfuse within restrictive guidelines or not. The handover of care between anesthesia providers has been linked with increased mortality,26 yet the handover of patient care during surgery both by in-room staff or attending did not appear to be a factor in compliance here. However, we did not specifically evaluate the timing of handovers relative to the timing of transfusion which limits the robustness of this conclusion.

While >70% anesthesia providers claim that they intended to follow a restrictive transfusion strategy, as implied by the selection of a transfusion target of hematocrit ≤24% within the transfusion tool, in only 30% of the cases was this threshold met. In the other 70% of transfusions, the hematocrit was >24% before transfusion. The most common reason for the selection of a relatively high hematocrit target range was ischemic heart disease, but the selection of elements from a pick list limits the choice of factors that truly and completely influence transfusion practice.

There are a number of limitations with this study. The study design was retrospective, standardized care was not dictated and there was no concurrent control group. Measurement of hematocrit required an arterial or venous sample of blood; the presence of invasive monitoring would have made sampling easier and may have been an unmeasured confounder. Use of the transfusion tool was not made mandatory; clinicians may have selected to use the tool differentially depending on their perceived ability to be compliant yet hematocrit checking did not appear to vary between anesthesiology attendings. We cannot ascribe causality to any of the independent associations with transfusion practice, and it is difficult to guarantee that the multiple causal factors that determine background change were adequately controlled for within our models.23 Indeed, subtle concepts of appropriateness probably constantly evolve.27 The fact that the behaviors undergoing evaluation here had plateaued within the period after the introduction of the transfusion guideline makes the impact of any such background shift less likely and makes alternative methodology such as the difference-in-differences less applicable28; identifying a control group would also be extremely challenging. We cannot exclude change (related to intraoperative practice or not) in transfusion behavior occurring outside of the operating room either preoperatively or postoperatively that may have impacted safety outcomes. The primary target behavior of checking a hematocrit is not controversial, but it is possible that anesthesia providers were unaware of centrally available laboratory values that may have confounded the results. It is also possible that clinicians made rational transfusion decisions based on the latest hematocrit and subsequent blood loss but were measured noncompliant by our definitions. It is possible that some patients bled rapidly and/or displayed hemodynamic instability making transfusion decisions clinically appropriate despite an above transfusion threshold hematocrit because hematocrit is not the sole potential trigger for transfusion. The evaluation of hemodynamic data on an individual patient basis was beyond the scope of the study, and given the relatively low EBL, such cases likely did not make up a large proportion of the subjects. The evidence supporting restrictive transfusion is expansive and widely accepted29–33 and such a strategy depends on the practitioner knowing the hematocrit. It is important to note that subsets of surgical patients exist who may benefit from a more liberal transfusion strategy,34–37 notably high-risk patients undergoing major surgery25; when controversy exists, the predetermination of transfusion thresholds is likely more important than ever. This work was undertaken at a single large academic center. It is debatable whether or not the described system could be implemented within other health care organizations; however, the principles of discussing a transfusion target with the surgical team and agreeing transfusion thresholds before surgery are simple, may be helpful and do not require software support.

In conclusion, the introduction of a hospital-wide transfusion guideline was independently associated with increased intraoperative pretransfusion hematocrit assessment and restrictive transfusion. The use of a novel intraoperative software transfusion tool was not associated with pretransfusion hematocrit checking or the adoption of a restrictive transfusion strategy within the operating room.

Back to Top | Article Outline

DISCLOSURES

Name: Paul Picton, MBChB, MRCP, FRCA.

Contribution: This author helped conceive and design the study, and prepared all versions of the manuscript.

Name: Jordan Starr, MD.

Contribution: This author helped conduct the study analytics.

Name: Sachin Kheterpal, MD, MBA.

Contribution: This author helped conceive and design the study.

Name: Aleda M. L. Thompson, MS.

Contribution: This author helped conduct the study analytics.

Name: Michelle Housey, MPH.

Contribution: This author helped conduct the study analytics.

Name: Subramanian Sathishkumar, MD, FRCA.

Contribution: This author helped provide critical appraisal of the study.

Name: Timur Dubovoy, MD.

Contribution: This author helped provide critical appraisal of the study.

Name: Nathan Kirkpatrick, BA.

Contribution: This author helped conduct the study analytics.

Name: Kevin K. Tremper, MD, PhD.

Contribution: This author helped conceive and design the study.

Name: Milo Engoren, MD.

Contribution: This author helped conceive and design the study, and provide the critical appraisal of the study.

Name: Satya Krishna Ramachandran, MD.

Contribution: This author helped conceive and design the study, and provide the critical appraisal of the study.

This manuscript was handled by: Marisa B. Marques, MD.

Back to Top | Article Outline

REFERENCES

1. Whitaker B, Rajbhandary S, Kleinman S, Harris A, Kamani N. Trends in United States blood collection and transfusion: results from the 2013 AABB Blood Collection, Utilization, and Patient Blood Management Survey. Transfusion. 2016;56:2173–2183.
2. Whitaker B, Schlumpf K, Schulman J, Green J. The 2009 National Blood Collection and Utilization Survey Report. 2011.Washington, DC: US Department of Health and Human Services.
3. Wells AW, Mounter PJ, Chapman CE, Stainsby D, Wallis JP. Where does blood go? Prospective observational study of red cell transfusion in North England. BMJ. 2002;325:803.
4. Carson JL, Noveck H, Berlin JA, Gould SA. Mortality and morbidity in patients with very low postoperative Hb levels who decline blood transfusion. Transfusion. 2002;42:812–818.
5. Marik PE, Corwin HL. Efficacy of red blood cell transfusion in the critically ill: a systematic review of the literature. Crit Care Med. 2008;36:2667–2674.
6. Hébert PC, Wells G, Blajchman MA, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. Transfusion requirements in critical care investigators, Canadian Critical Care Trials Group. N Engl J Med. 1999;340:409–417.
7. Lacroix J, Hébert PC, Hutchison JS, et al; TRIPICU Investigators; Canadian Critical Care Trials Group; Pediatric Acute Lung Injury and Sepsis Investigators Network. Transfusion strategies for patients in pediatric intensive care units. N Engl J Med. 2007;356:1609–1619.
8. Hajjar LA, Vincent JL, Galas FR, et al. Transfusion requirements after cardiac surgery: the TRACS randomized controlled trial. JAMA. 2010;304:1559–1567.
9. Carson JL, Terrin ML, Noveck H, et al; FOCUS Investigators. Liberal or restrictive transfusion in high-risk patients after hip surgery. N Engl J Med. 2011;365:2453–2462.
10. Villanueva C, Colomo A, Bosch A, et al. Transfusion strategies for acute upper gastrointestinal bleeding. N Engl J Med. 2013;368:11–21.
11. Goodnough LT, Shieh L, Hadhazy E, Cheng N, Khari P, Maggio P. Improved blood utilization using real-time clinical decision support. Transfusion. 2014;54:1358–1365.
12. Zuckerberg GS, Scott AV, Wasey JO, et al. Efficacy of education followed by computerized provider order entry with clinician decision support to reduce red blood cell utilization. Transfusion. 2015;55:1628–1636.
13. Carson JL, Guyatt G, Heddle NM, et al. Clinical practice guidelines from the AABB: red blood cell transfusion thresholds and storage. JAMA. 2016;316:2025–2035.
14. Houle TT. Importance of effect sizes for the accumulation of knowledge. Anesthesiology. 2007;106:415–417.
15. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143:29–36.
16. Harrell F Jr.. Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. 2001:New York, NY: Springer87–103.
17. Sathishkumar S, Lai M, Picton P, et al. Behavioral modification of intraoperative hyperglycemia management with a novel real-time audiovisual monitor. Anesthesiology. 2015;123:29–37.
18. Ehrenfeld JM, Wanderer JP, Terekhov M, Rothman BS, Sandberg WS. A perioperative systems design to improve intraoperative glucose monitoring is associated with a reduction in surgical site infections in a diabetic patient population. Anesthesiology. 2017;126:431–440.
19. Blum JM, Stentz MJ, Maile MD, et al. Automated alerting and recommendations for the management of patients with preexisting hypoxia and potential acute lung injury: a pilot study. Anesthesiology. 2013;119:295–302.
20. Ehrenfeld JM, Epstein RH, Bader S, Kheterpal S, Sandberg WS. Automatic notifications mediated by anesthesia information management systems reduce the frequency of prolonged gaps in blood pressure documentation. Anesth Analg. 2011;113:356–363.
21. Lowe A, Jones RW, Harrison MJ. The graphical presentation of decision support information in an intelligent anaesthesia monitor. Artif Intell Med. 2001;22:173–191.
22. Goodnough LT, Shah N. Is there a “magic” hemoglobin number? Clinical decision support promoting restrictive blood transfusion practices. Am J Hematol. 2015;90:927–933.
23. Dimick JB, Ryan AM. Methods for evaluating changes in health care policy: the difference-in-differences approach. JAMA. 2014;312:2401–2402.
24. Holst LB, Haase N, Wetterslev J, et al; TRISS Trial Group; Scandinavian Critical Care Trials Group. Lower versus higher hemoglobin threshold for transfusion in septic shock. N Engl J Med. 2014;371:1381–1391.
25. Hovaguimian F, Myles PS. Restrictive versus liberal transfusion strategy in the perioperative and acute care settings: a context-specific systematic review and meta-analysis of randomized controlled trials. Anesthesiology. 2016;125:46–61.
26. Hudson CC, McDonald B, Hudson JK, Tran D, Boodhwani M. Impact of anesthetic handover on mortality and morbidity in cardiac surgery: a cohort study. J Cardiothorac Vasc Anesth. 2015;29:11–16.
27. Liao KF, Lin CL, Lai SW, Chen WC. Zolpidem use associated with increased risk of pyogenic liver abscess: a case-control study in Taiwan. Medicine (Baltimore). 2015;94:e1302.
28. Ryan AM, Burgess JF Jr, Dimick JB. Why we should not be indifferent to specification choices for difference-in-differences. Health Serv Res. 2015;50:1211–1235.
29. American Society of Anesthesiologists Task Force on Perioperative Blood Management. Practice guidelines for perioperative blood management: an updated report by the American Society of Anesthesiologists Task Force on Perioperative Blood Management. Anesthesiology. 2015;122:241–275.
30. Carson JL, Carless PA, Hebert PC. Transfusion thresholds and other strategies for guiding allogeneic red blood cell transfusion. Cochrane Database Syst Rev. 2012;4:CD002042.
31. Curley GF, Shehata N, Mazer CD, Hare GM, Friedrich JO. Transfusion triggers for guiding RBC transfusion for cardiovascular surgery: a systematic review and meta-analysis. Crit Care Med. 2014;42:2611–2624.
32. Holst LB, Petersen MW, Haase N, Perner A, Wetterslev J. Restrictive versus liberal transfusion strategy for red blood cell transfusion: systematic review of randomised trials with meta-analysis and trial sequential analysis. BMJ. 2015;350:h1354.
33. Salpeter SR, Buckley JS, Chatterjee S. Impact of more restrictive blood transfusion strategies on clinical outcomes: a meta-analysis and systematic review. Am J Med. 2014;127:124–131.e3.
34. Murphy GJ, Pike K, Rogers CA, et al; TITRe2 Investigators. Liberal or restrictive transfusion after cardiac surgery. N Engl J Med. 2015;372:997–1008.
35. Wu WC, Smith TS, Henderson WG, et al. Operative blood loss, blood transfusion, and 30-day mortality in older patients after major noncardiac surgery. Ann Surg. 2010;252:11–17.
36. Wu WC, Trivedi A, Friedmann PD, et al. Association between hospital intraoperative blood transfusion practices for surgical blood loss and hospital surgical mortality rates. Ann Surg. 2012;255:708–714.
37. Fominskiy E, Putzu A, Monaco F, et al. Liberal transfusion strategy improves survival in perioperative but not in critically ill patients. A meta-analysis of randomised trials. Br J Anaesth. 2015;115:511–519.
© 2018 International Anesthesia Research Society