- Email: [email protected]
Restrictive red blood cell transfusion: not just for the stable intensive care unit patient
The American Journal of Surgery (2008) 195, 803– 806
Clinical Surgery—American
Restrictive red blood cell transfusion: not just for the stable intensive care unit patient Wendy L. Wahl, M.D.a,*, Mark R. Hemmila, M.D.a, Paul M. Maggio, M.D., M.B.A.a, Saman Arbabi, M.D., M.P.H.b a
Trauma Burn Center, University of Michigan Health System, 1500 E. Medical Center Drive, Ann Arbor, MI 48190-5539, USA; bUniversity of Washington Harborview Medical Center, Seattle, WA, USA KEYWORDS: Transfusion; Guideline; Red blood cells; Trauma; Burn; Intensive care unit
Abstract BACKGROUND: Multiple studies report that patients receiving red blood cell (RBC) transfusion in the intensive care unit (ICU) are more likely to experience complications. Despite these findings, surgical patients are frequently transfused for operative procedures, trauma, and burns. We hypothesized that a RBC transfusion guideline would safely decrease our use of RBC transfusions in the ICU and lower the hematocrit at which our trauma and burn patients were transfused, both in the stable and symptomatic patient. METHODS: For each episode of RBC transfusion, the pretransfusion vital signs and reasons for transfusion were recorded prospectively from August 2003 through April 2004. Before institution of the transfusion guideline, which stressed withholding transfusion for hematocrit over 23 in asymptomatic patients, intensive education of all caregivers occurred. Data from all transfusions during 2005 were also reviewed for long-term compliance with the guideline. RESULTS: Eighty-two of 316 ICU patients (26%) had 315 RBC transfusion events during the initial study period. Mean transfusion hematocrits decreased from 26.6 ⫾ 4.7 to 23.9 ⫾ 2.6 (P ⬍ .0003) for all patients. For the follow-up period in 2005, 94 of 523 patients (18%) were transfused in the ICU at a mean transfusion hematocrit of 24.1 for symptomatic (P ⬍ .0001) and 22.5 for asymptomatic patients (P ⬍ .0001). Low hematocrit was the most frequently cited reason for transfusion for all patients in the first part of the study, whereas hemodynamic instability (n ⫽ 91 events) and perioperative losses (n ⫽ 49 events) ranked highest for symptomatic patients. CONCLUSION: A transfusion guideline accompanied by intensive education is effective in reducing RBC transfusions in a trauma-burn ICU. A lower hematocrit was well tolerated in both the symptomatic and asymptomatic groups of surgical patients. With education and follow-up, the changes in transfusion practices were durable and affected transfusion practices for both asymptomatic and symptomatic patients. © 2008 Elsevier Inc. All rights reserved.
* Corresponding author. Tel.: ⫹1-734-936-9666; fax: ⫹1-734-9369657. E-mail address: [email protected] Manuscript received January 24, 2007; revised manuscript May 22, 2007
0002-9610/$ - see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.amjsurg.2007.05.047
Multiple studies have now reported that blood product transfusion can negatively impact patient outcomes in the intensive care unit (ICU).1–3 Decreased transfusion of red blood cell (RBC) products to maintain hemoglobin at a much lower than the traditional transfusion threshold of 10 g/dL appears to be safe and tolerated by most ICU patient populations studied.1 In addition to overall tolerance of a
804
The American Journal of Surgery, Vol 195, No 6, June 2008
Methods
Figure 1 Trauma-burn ICU patient blood use. HR, heart rate; BP, blood pressure; Hgb, hemoglobin; Hct, hematocrit; ABG, arterial blood gas; CAD, coronary artery disease; OR, operating room.
lower transfusion trigger, ICU patients exhibited lower infection rates, mortality, and hospital length of stay when given fewer transfusions.1,2 In both the Canadian Transfusion Requirements in Critical Care and American CRIT studies, mixed populations of patients were included, with a preponderance of medical patients. Recently, prospective studies in trauma patients have shown that RBC transfusion was an independent predictor of subsequent infections and death.3–5 Similar findings have now been reported for the burn population.6 Both trauma and burn patients are at risk for periods of acute blood loss anemia related to their injuries and operative interventions. Reluctance to withhold RBC transfusion in trauma and burn patients may be related to the concern over ongoing bleeding from injuries and surgical interventions. After these initial episodes of blood loss, trauma and burn patients become more like medical ICU patients, yet the initial bias may still lead caregivers to transfuse. Because of this potential bias, guidelines for transfusion in the trauma patient were recently proposed by the investigators of the “Surgical Glue Grant” (Inflammation and the Host Response to Injury, a Large-Scale Collaborative Project) for withholding RBC transfusion for patients with a hemoglobin ⬎7 g/dL (or a hematocrit ⬎21%) unless the patient had severe cardiovascular disease. These recommendations were for patients who were considered stable in the ICU without ongoing bleeding.7 The purpose of this study was to examine the effect of our RBC transfusion guideline on RBC transfusion practices in our trauma-burn ICU. We hypothesized that with the use of the proposed guideline the pretransfusion hematocrit would be lower for all patients without adverse sequelae.
Data on transfusions for all patients admitted to our 10-bed trauma-burn ICU (TBICU) during a 9-month period (August 2003 through April 2004) was prospectively collected. Immediately before the study period, physicians, nurses, and ancillary care providers in the TBICU were educated about the new RBC transfusion guideline. During this time, we provided monthly education of house officers rotating in the TBICU. Periodic reminders during weekly multidisciplinary rounds were issued about the new guideline. The initial study period is designated “2003– 4.” We compared the first 3 months of 2003– 4, when education was ongoing, to the last 3 months of 2003– 4, after the guideline had been in use. Our guideline recommended withholding RBC transfusion for any hematocrit of 23 or above unless patients were symptomatic. Symptoms included hypotension, use of pressors for blood pressure support, poor oxygenation despite maximization of ventilator mechanics, active bleeding from injuries or a gastrointestinal source, angina, or evidence of acute myocardial ischemia. Transfusions were recorded for all patients during their entire ICU stay. Packed RBCs given during surgical procedures in the operating room were not examined. For 2003– 4, data were collected for each episode of RBC transfusion (1 set of data, even if more than 1 U was ordered for transfusion), including vital signs, hematocrit, reasons for transfusion, and blood gas measurements (if available). Data forms were completed by the bedside nurse (Fig. 1). Patient data including age, history of coronary artery disease, Injury Severity Score for trauma patients, acute physiology and chronic health evaluation III (APACHE III) scores, mortality, ICU and hospital lengths of stay, and number of RBC transfusions were collected. All RBC transfusions for 2005 were evaluated for the presence or absence of symptoms at the time of transfusion and for the pretransfusion hematocrit. Reminders of the unit protocol were given to surgical housestaff on rounds and at the
Figure 2
The mean hematocrit at the time of transfusion.
W.L. Wahl et al. Table 1
Restrictive red blood cell transfusion
805
Demographics and outcomes for transfused patients
Time period
n (%) Transfused/all patients
APACHE III
Age (y)
Mortality n (%)
ICU LOS (d)
Hospital LOS (d)
2003–4 2005 P value (unadjusted) P value (adjusted*)
82/316 (26) 94/523 (18)
41 ⫾ 24 43 ⫾ 26 ⬍0.001
43 ⫾ 22 45 ⫾ 21 0.4
33 (10) 46 (9) 0.4 0.06
6 ⫾ 10 5⫾9 0.5 0.3
15 ⫾ 22 13 ⫾ 15 0.05 0.02
*Adjusted for age and APACHE III.
start of the unit rotation, but no further formal education was provided regarding transfusion. Chi-square analysis for dichotomous values, Student t tests for continuous variables, and Mantel-Haenszel incidence rate analysis for percentage values were performed. Mortality and length of stay were also evaluated by means of multivariate logistic and linear regression, after adjusting for APACHE III and age. Stata Statistics/Data Analysis 8.0 software (Stata Corp, College Station, TX) was used. The institutional review board approved this study.
not for the heart rate. The mean systolic blood pressure was 105 ⫾ 20 compared with 123 ⫾ 22 mm Hg (P ⬍ .001), the urine output was 36 ⫾ 52 versus 82 ⫾ 105 mL/h (P ⬍ .001), and the heart rate was 104 ⫾ 22 compared with 105 ⫾ 22 beats/min (not significant) for the symptomatic and asymptomatic groups, respectively. To evaluate the durability of this guideline, the RBC transfusion data from 2005 were collected. Ninety-four of 523 patients (18%) received RBC transfusions in 2005. For the same time period, the pretransfusion hematocrit values of 24.1 ⫾ 3 for symptomatic and 22.5 ⫾ 2 for asymptomatic patients were significantly lower than 2003– 4 (Table 3). The baseline pretransfusion hematocrit before 2003 was 26.8 for all transfused patients. Comparing 2003– 4 data with 2005, there was no significant difference in age (Table 1). However, admission APACHE III scores were significantly higher for 2005 patients (Table 1). Multivariate regression analysis, adjusting for age and APACHE III, showed a trend toward lower mortality in 2005 patients (odds ratio, 0.6; P ⫽ .06). There was a significant decrease in the hospital length of stay for 2005 patients, and the mean adjusted difference in hospital length of stay was 3 days less for 2005 (P ⫽ .02).
Results Of 316 patients admitted during the initial 9-month period, 82 (26%) patients had 315 RBC transfusion events. Data forms were completed for 83% of transfusion events. The overall pretransfusion hematocrit decreased from the first 3 months to the last 3 months of 2003– 4, from 26.6 ⫾ 4.7 to 23.9 ⫾ 2.2 (P ⬍ .001) (Fig. 2). The percentage of patients who were considered symptomatic at the time of transfusion increased from 57% to 83% (P ⬍ .05), whereas the overall percentage of patients transfused with a hematocrit over 23 decreased from 43% to 18% (P ⬍ .05) during 2003– 4. Although the percent of patients receiving transfusions decreased and the overall pretransfusion hematocrit dropped in 2003– 4, the overall acuity and age of the patients remained fairly constant (Table 1). The differences in mortality and ICU length of stay did not reach statistical significance (Table 1). The reasons for transfusion are listed for the symptomatic compared with the asymptomatic groups for 2003– 4 (Table 2). The vital signs pretransfusion for the symptomatic versus the asymptomatic group were statistically different for systolic blood pressure and urine output per hour but
Table 2
Comments Similar to other ICU populations, a high percentage of trauma and burn patients were transfused at the beginning of the study period. Over time, this percentage decreased, as did the mean pretransfusion hematocrit and the number of asymptomatic patients transfused for a low hematocrit. The effect of the transfusion guideline was not limited to asymptomatic patients receiving RBC transfusion but extended to “symptomatic” patients as well. The lower mean transfusion hematocrit for the symptomatic patients in the
Reasons cited for transfusion of RBCs (more than one per patient)
Number of reasons cited for transfusion events
Unstable/pressors
Oxygenation
PMH CAD
Active Angina
Dyspnea
Perioperative
Active bleeding
Low hematocrit
Asymptomatic (n ⫽ 122) Symptomatic (n ⫽ 177)
0 91
8 29
8 16
0 0
1 7
32 49
8 31
100 107
PMH ⫽ past medical history; CAD ⫽ coronary artery disease.
806 Table 3 event
2003–4 2005 P value
The American Journal of Surgery, Vol 195, No 6, June 2008 Last documented hematocrit preceding transfusion Asymptomatic patients
Symptomatic patients
n (%)
Mean Hct
n (%)
Mean Hct
125 (40) 43 (46)
24.9 ⫾ 3 22.5 ⫾ 2 ⬍.0001
190 (60) 51 (54)
26.8 ⫾ 4 24.1 ⫾ 3 ⬍.0001
follow-up part of the study suggests that care providers became more comfortable with deferring transfusions for critically ill patients on ventilators receiving pressor medications or in the acute postinjury phase despite a low hematocrit. We did not exclude patients who were being actively resuscitated or were in the postoperative period. The decrease in the transfusion trigger and the resulting lower pretransfusion hematocrit did not negatively impact the mortality or length of stay, supporting our theory that restricted transfusion in trauma and burn patients is safe overall. The reasons cited for transfusion suggest that a scheduled operative procedure prompts physicians to transfuse for a low hematocrit even in asymptomatic patients. These data may be useful in targeting future avenues to impact transfusion practices among surgical patients with delay of transfusions until symptoms or until the operative procedure is underway and significant blood loss occurs. The transfusion trigger of hematocrit less than 23 was initially chosen to compromise between the traditional number of 30 and the transfusion trigger of 21.5 in the Transfusion Requirements in Critical Care study because of our concerns of potential bleeding in fresh postoperative or postinjury patients. The threshold of 23 as the hematocrit transfusion trigger was also lower than our previous mean pretransfusion hematocrit of 26.8. With data that a transfusion trigger of 23 did not adversely affect mortality and length of stay, there is potential opportunity to decrease RBC transfusions further. It is interesting that although the mean pretransfusion hematocrit for 2005 was lower than 2003– 4, the percent of “symptomatic” patients in 2005 was lower at 54% compared with 60% during 2003– 4. These data would suggest that lowering the protocol hematocrit transfusion trigger may drive the decision to transfuse to an even lower level. Poor oxygenation was listed as a reason for transfusion in 29 patients in the symptomatic group, yet we have no data that transfusion improved the condition of these patients. This study was not designed to detect improvements in oxygenation or pulmonary function after transfusion.
Although this study did not show a clear improvement in mortality, we did find a shorter hospital length of stay similar to other authors studying transfused trauma and burn patients.3– 6 The small number of patients with a limited sample size and heterogenous population may have obscured the survival benefit, although there was a trend toward improved outcomes. With the follow-up data that the overall pretransfusion hematocrit remained low even after the study period, it appears that our guideline, after the initial intensive education period, has become our routine practice. The lower hematocrit at the time of transfusion for “symptomatic” patients would imply that physicians are more comfortable withholding transfusion in critically ill surgical patients for reasons other than active bleeding. Although we have seen a small improvement in mortality and length of stay in the year follow-up period, other changes in our guidelines and standards of care may have contributed to improved outcomes such as tighter glucose control and use of beta-blockers in more patients. In summary, we found that a RBC transfusion guideline was effective in decreasing the overall number of RBC transfusions in both symptomatic and asymptomatic trauma and burn patients. The number of trauma and burn patients transfused for the sole indication of an asymptomatic but low hematocrit decreased. More restrictive transfusion practices did not worsen survival and are now part of our standard practice.
References 1. Hebert PC, Wells G, Martin C, et al. A Canadian survey of transfusion practices in critically ill patients. Crit Care Med 1998;26:482–7. 2. Corwin HL, Gettinger A, Pearl RG, et al. The CRIT Study: anemia and blood transfusion in the critically ill. Current clinical practice in the United States. Crit Care Med 2004;32;29 –52. 3. Malone DL, Dunne J, Tracy JK, et al. Blood transfusion, independent of shock severity, is associated with worse outcome in trauma. J Trauma 2003;54:898 –905. 4. Shapiro MJ, Gettinger A, Corwin HL, et al. Anemia and blood transfusion in trauma patients admitted to the intensive care unit. J Trauma 2003;55:269 –73. 5. Croce MA, Tolley EA, Claridge JA, et al. Transfusions result in pulmonary morbidity and death after a moderate degree of injury. J Trauma 2005;59:19 –23. 6. Palmieri TL, Caruso DM, Foster KN, et al. Effect of blood transfusion on outcome after major burn injury: a multicenter study. Crit Care Med 2006;34:1602–7. 7. West MA, Shapiro MB, Nathens AB, et al. Inflammation and the host response to injury, a large-scale collaborative project: patient-oriented research core-standard operating procedures for clinical care. IV Guidelines for transfusion in the trauma patient. J Trauma 2006;61:436 –9.
Clinical Surgery—American
Restrictive red blood cell transfusion: not just for the stable intensive care unit patient Wendy L. Wahl, M.D.a,*, Mark R. Hemmila, M.D.a, Paul M. Maggio, M.D., M.B.A.a, Saman Arbabi, M.D., M.P.H.b a
Trauma Burn Center, University of Michigan Health System, 1500 E. Medical Center Drive, Ann Arbor, MI 48190-5539, USA; bUniversity of Washington Harborview Medical Center, Seattle, WA, USA KEYWORDS: Transfusion; Guideline; Red blood cells; Trauma; Burn; Intensive care unit
Abstract BACKGROUND: Multiple studies report that patients receiving red blood cell (RBC) transfusion in the intensive care unit (ICU) are more likely to experience complications. Despite these findings, surgical patients are frequently transfused for operative procedures, trauma, and burns. We hypothesized that a RBC transfusion guideline would safely decrease our use of RBC transfusions in the ICU and lower the hematocrit at which our trauma and burn patients were transfused, both in the stable and symptomatic patient. METHODS: For each episode of RBC transfusion, the pretransfusion vital signs and reasons for transfusion were recorded prospectively from August 2003 through April 2004. Before institution of the transfusion guideline, which stressed withholding transfusion for hematocrit over 23 in asymptomatic patients, intensive education of all caregivers occurred. Data from all transfusions during 2005 were also reviewed for long-term compliance with the guideline. RESULTS: Eighty-two of 316 ICU patients (26%) had 315 RBC transfusion events during the initial study period. Mean transfusion hematocrits decreased from 26.6 ⫾ 4.7 to 23.9 ⫾ 2.6 (P ⬍ .0003) for all patients. For the follow-up period in 2005, 94 of 523 patients (18%) were transfused in the ICU at a mean transfusion hematocrit of 24.1 for symptomatic (P ⬍ .0001) and 22.5 for asymptomatic patients (P ⬍ .0001). Low hematocrit was the most frequently cited reason for transfusion for all patients in the first part of the study, whereas hemodynamic instability (n ⫽ 91 events) and perioperative losses (n ⫽ 49 events) ranked highest for symptomatic patients. CONCLUSION: A transfusion guideline accompanied by intensive education is effective in reducing RBC transfusions in a trauma-burn ICU. A lower hematocrit was well tolerated in both the symptomatic and asymptomatic groups of surgical patients. With education and follow-up, the changes in transfusion practices were durable and affected transfusion practices for both asymptomatic and symptomatic patients. © 2008 Elsevier Inc. All rights reserved.
* Corresponding author. Tel.: ⫹1-734-936-9666; fax: ⫹1-734-9369657. E-mail address: [email protected] Manuscript received January 24, 2007; revised manuscript May 22, 2007
0002-9610/$ - see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.amjsurg.2007.05.047
Multiple studies have now reported that blood product transfusion can negatively impact patient outcomes in the intensive care unit (ICU).1–3 Decreased transfusion of red blood cell (RBC) products to maintain hemoglobin at a much lower than the traditional transfusion threshold of 10 g/dL appears to be safe and tolerated by most ICU patient populations studied.1 In addition to overall tolerance of a
804
The American Journal of Surgery, Vol 195, No 6, June 2008
Methods
Figure 1 Trauma-burn ICU patient blood use. HR, heart rate; BP, blood pressure; Hgb, hemoglobin; Hct, hematocrit; ABG, arterial blood gas; CAD, coronary artery disease; OR, operating room.
lower transfusion trigger, ICU patients exhibited lower infection rates, mortality, and hospital length of stay when given fewer transfusions.1,2 In both the Canadian Transfusion Requirements in Critical Care and American CRIT studies, mixed populations of patients were included, with a preponderance of medical patients. Recently, prospective studies in trauma patients have shown that RBC transfusion was an independent predictor of subsequent infections and death.3–5 Similar findings have now been reported for the burn population.6 Both trauma and burn patients are at risk for periods of acute blood loss anemia related to their injuries and operative interventions. Reluctance to withhold RBC transfusion in trauma and burn patients may be related to the concern over ongoing bleeding from injuries and surgical interventions. After these initial episodes of blood loss, trauma and burn patients become more like medical ICU patients, yet the initial bias may still lead caregivers to transfuse. Because of this potential bias, guidelines for transfusion in the trauma patient were recently proposed by the investigators of the “Surgical Glue Grant” (Inflammation and the Host Response to Injury, a Large-Scale Collaborative Project) for withholding RBC transfusion for patients with a hemoglobin ⬎7 g/dL (or a hematocrit ⬎21%) unless the patient had severe cardiovascular disease. These recommendations were for patients who were considered stable in the ICU without ongoing bleeding.7 The purpose of this study was to examine the effect of our RBC transfusion guideline on RBC transfusion practices in our trauma-burn ICU. We hypothesized that with the use of the proposed guideline the pretransfusion hematocrit would be lower for all patients without adverse sequelae.
Data on transfusions for all patients admitted to our 10-bed trauma-burn ICU (TBICU) during a 9-month period (August 2003 through April 2004) was prospectively collected. Immediately before the study period, physicians, nurses, and ancillary care providers in the TBICU were educated about the new RBC transfusion guideline. During this time, we provided monthly education of house officers rotating in the TBICU. Periodic reminders during weekly multidisciplinary rounds were issued about the new guideline. The initial study period is designated “2003– 4.” We compared the first 3 months of 2003– 4, when education was ongoing, to the last 3 months of 2003– 4, after the guideline had been in use. Our guideline recommended withholding RBC transfusion for any hematocrit of 23 or above unless patients were symptomatic. Symptoms included hypotension, use of pressors for blood pressure support, poor oxygenation despite maximization of ventilator mechanics, active bleeding from injuries or a gastrointestinal source, angina, or evidence of acute myocardial ischemia. Transfusions were recorded for all patients during their entire ICU stay. Packed RBCs given during surgical procedures in the operating room were not examined. For 2003– 4, data were collected for each episode of RBC transfusion (1 set of data, even if more than 1 U was ordered for transfusion), including vital signs, hematocrit, reasons for transfusion, and blood gas measurements (if available). Data forms were completed by the bedside nurse (Fig. 1). Patient data including age, history of coronary artery disease, Injury Severity Score for trauma patients, acute physiology and chronic health evaluation III (APACHE III) scores, mortality, ICU and hospital lengths of stay, and number of RBC transfusions were collected. All RBC transfusions for 2005 were evaluated for the presence or absence of symptoms at the time of transfusion and for the pretransfusion hematocrit. Reminders of the unit protocol were given to surgical housestaff on rounds and at the
Figure 2
The mean hematocrit at the time of transfusion.
W.L. Wahl et al. Table 1
Restrictive red blood cell transfusion
805
Demographics and outcomes for transfused patients
Time period
n (%) Transfused/all patients
APACHE III
Age (y)
Mortality n (%)
ICU LOS (d)
Hospital LOS (d)
2003–4 2005 P value (unadjusted) P value (adjusted*)
82/316 (26) 94/523 (18)
41 ⫾ 24 43 ⫾ 26 ⬍0.001
43 ⫾ 22 45 ⫾ 21 0.4
33 (10) 46 (9) 0.4 0.06
6 ⫾ 10 5⫾9 0.5 0.3
15 ⫾ 22 13 ⫾ 15 0.05 0.02
*Adjusted for age and APACHE III.
start of the unit rotation, but no further formal education was provided regarding transfusion. Chi-square analysis for dichotomous values, Student t tests for continuous variables, and Mantel-Haenszel incidence rate analysis for percentage values were performed. Mortality and length of stay were also evaluated by means of multivariate logistic and linear regression, after adjusting for APACHE III and age. Stata Statistics/Data Analysis 8.0 software (Stata Corp, College Station, TX) was used. The institutional review board approved this study.
not for the heart rate. The mean systolic blood pressure was 105 ⫾ 20 compared with 123 ⫾ 22 mm Hg (P ⬍ .001), the urine output was 36 ⫾ 52 versus 82 ⫾ 105 mL/h (P ⬍ .001), and the heart rate was 104 ⫾ 22 compared with 105 ⫾ 22 beats/min (not significant) for the symptomatic and asymptomatic groups, respectively. To evaluate the durability of this guideline, the RBC transfusion data from 2005 were collected. Ninety-four of 523 patients (18%) received RBC transfusions in 2005. For the same time period, the pretransfusion hematocrit values of 24.1 ⫾ 3 for symptomatic and 22.5 ⫾ 2 for asymptomatic patients were significantly lower than 2003– 4 (Table 3). The baseline pretransfusion hematocrit before 2003 was 26.8 for all transfused patients. Comparing 2003– 4 data with 2005, there was no significant difference in age (Table 1). However, admission APACHE III scores were significantly higher for 2005 patients (Table 1). Multivariate regression analysis, adjusting for age and APACHE III, showed a trend toward lower mortality in 2005 patients (odds ratio, 0.6; P ⫽ .06). There was a significant decrease in the hospital length of stay for 2005 patients, and the mean adjusted difference in hospital length of stay was 3 days less for 2005 (P ⫽ .02).
Results Of 316 patients admitted during the initial 9-month period, 82 (26%) patients had 315 RBC transfusion events. Data forms were completed for 83% of transfusion events. The overall pretransfusion hematocrit decreased from the first 3 months to the last 3 months of 2003– 4, from 26.6 ⫾ 4.7 to 23.9 ⫾ 2.2 (P ⬍ .001) (Fig. 2). The percentage of patients who were considered symptomatic at the time of transfusion increased from 57% to 83% (P ⬍ .05), whereas the overall percentage of patients transfused with a hematocrit over 23 decreased from 43% to 18% (P ⬍ .05) during 2003– 4. Although the percent of patients receiving transfusions decreased and the overall pretransfusion hematocrit dropped in 2003– 4, the overall acuity and age of the patients remained fairly constant (Table 1). The differences in mortality and ICU length of stay did not reach statistical significance (Table 1). The reasons for transfusion are listed for the symptomatic compared with the asymptomatic groups for 2003– 4 (Table 2). The vital signs pretransfusion for the symptomatic versus the asymptomatic group were statistically different for systolic blood pressure and urine output per hour but
Table 2
Comments Similar to other ICU populations, a high percentage of trauma and burn patients were transfused at the beginning of the study period. Over time, this percentage decreased, as did the mean pretransfusion hematocrit and the number of asymptomatic patients transfused for a low hematocrit. The effect of the transfusion guideline was not limited to asymptomatic patients receiving RBC transfusion but extended to “symptomatic” patients as well. The lower mean transfusion hematocrit for the symptomatic patients in the
Reasons cited for transfusion of RBCs (more than one per patient)
Number of reasons cited for transfusion events
Unstable/pressors
Oxygenation
PMH CAD
Active Angina
Dyspnea
Perioperative
Active bleeding
Low hematocrit
Asymptomatic (n ⫽ 122) Symptomatic (n ⫽ 177)
0 91
8 29
8 16
0 0
1 7
32 49
8 31
100 107
PMH ⫽ past medical history; CAD ⫽ coronary artery disease.
806 Table 3 event
2003–4 2005 P value
The American Journal of Surgery, Vol 195, No 6, June 2008 Last documented hematocrit preceding transfusion Asymptomatic patients
Symptomatic patients
n (%)
Mean Hct
n (%)
Mean Hct
125 (40) 43 (46)
24.9 ⫾ 3 22.5 ⫾ 2 ⬍.0001
190 (60) 51 (54)
26.8 ⫾ 4 24.1 ⫾ 3 ⬍.0001
follow-up part of the study suggests that care providers became more comfortable with deferring transfusions for critically ill patients on ventilators receiving pressor medications or in the acute postinjury phase despite a low hematocrit. We did not exclude patients who were being actively resuscitated or were in the postoperative period. The decrease in the transfusion trigger and the resulting lower pretransfusion hematocrit did not negatively impact the mortality or length of stay, supporting our theory that restricted transfusion in trauma and burn patients is safe overall. The reasons cited for transfusion suggest that a scheduled operative procedure prompts physicians to transfuse for a low hematocrit even in asymptomatic patients. These data may be useful in targeting future avenues to impact transfusion practices among surgical patients with delay of transfusions until symptoms or until the operative procedure is underway and significant blood loss occurs. The transfusion trigger of hematocrit less than 23 was initially chosen to compromise between the traditional number of 30 and the transfusion trigger of 21.5 in the Transfusion Requirements in Critical Care study because of our concerns of potential bleeding in fresh postoperative or postinjury patients. The threshold of 23 as the hematocrit transfusion trigger was also lower than our previous mean pretransfusion hematocrit of 26.8. With data that a transfusion trigger of 23 did not adversely affect mortality and length of stay, there is potential opportunity to decrease RBC transfusions further. It is interesting that although the mean pretransfusion hematocrit for 2005 was lower than 2003– 4, the percent of “symptomatic” patients in 2005 was lower at 54% compared with 60% during 2003– 4. These data would suggest that lowering the protocol hematocrit transfusion trigger may drive the decision to transfuse to an even lower level. Poor oxygenation was listed as a reason for transfusion in 29 patients in the symptomatic group, yet we have no data that transfusion improved the condition of these patients. This study was not designed to detect improvements in oxygenation or pulmonary function after transfusion.
Although this study did not show a clear improvement in mortality, we did find a shorter hospital length of stay similar to other authors studying transfused trauma and burn patients.3– 6 The small number of patients with a limited sample size and heterogenous population may have obscured the survival benefit, although there was a trend toward improved outcomes. With the follow-up data that the overall pretransfusion hematocrit remained low even after the study period, it appears that our guideline, after the initial intensive education period, has become our routine practice. The lower hematocrit at the time of transfusion for “symptomatic” patients would imply that physicians are more comfortable withholding transfusion in critically ill surgical patients for reasons other than active bleeding. Although we have seen a small improvement in mortality and length of stay in the year follow-up period, other changes in our guidelines and standards of care may have contributed to improved outcomes such as tighter glucose control and use of beta-blockers in more patients. In summary, we found that a RBC transfusion guideline was effective in decreasing the overall number of RBC transfusions in both symptomatic and asymptomatic trauma and burn patients. The number of trauma and burn patients transfused for the sole indication of an asymptomatic but low hematocrit decreased. More restrictive transfusion practices did not worsen survival and are now part of our standard practice.
References 1. Hebert PC, Wells G, Martin C, et al. A Canadian survey of transfusion practices in critically ill patients. Crit Care Med 1998;26:482–7. 2. Corwin HL, Gettinger A, Pearl RG, et al. The CRIT Study: anemia and blood transfusion in the critically ill. Current clinical practice in the United States. Crit Care Med 2004;32;29 –52. 3. Malone DL, Dunne J, Tracy JK, et al. Blood transfusion, independent of shock severity, is associated with worse outcome in trauma. J Trauma 2003;54:898 –905. 4. Shapiro MJ, Gettinger A, Corwin HL, et al. Anemia and blood transfusion in trauma patients admitted to the intensive care unit. J Trauma 2003;55:269 –73. 5. Croce MA, Tolley EA, Claridge JA, et al. Transfusions result in pulmonary morbidity and death after a moderate degree of injury. J Trauma 2005;59:19 –23. 6. Palmieri TL, Caruso DM, Foster KN, et al. Effect of blood transfusion on outcome after major burn injury: a multicenter study. Crit Care Med 2006;34:1602–7. 7. West MA, Shapiro MB, Nathens AB, et al. Inflammation and the host response to injury, a large-scale collaborative project: patient-oriented research core-standard operating procedures for clinical care. IV Guidelines for transfusion in the trauma patient. J Trauma 2006;61:436 –9.