- Email: [email protected]
Admission Hematocrit and Transfusion Requirements after Trauma
Admission Hematocrit and Transfusion Requirements after Trauma Chad M Thorson, MD, MSPH, Robert M Van Haren, MD, Mark L Ryan, MD, Reginald Pereira, Jeremy Olloqui, BS, Gerardo A Guarch, MD, Jose M Barrera, MD, Alexander M Busko, BS, Alan S Livingstone, MD, FACS, Kenneth G Proctor, PhD BACKGROUND:
METHODS:
RESULTS:
CONCLUSION:
BS,
The decision to transfuse packed RBCs (PRBC) during initial resuscitation of trauma patients is based on physiologic state, evidence for blood loss, and potential for ongoing hemorrhage. Initial hematocrit (Hct) is not considered an accurate marker of blood loss. This study tests the hypothesis that admission Hct is associated with transfusion requirements after trauma. From June to December 2008, data from 1,492 consecutive admissions at a Level I trauma center were retrospectively reviewed to determine whether initial Hct was associated with PRBC transfusions. From October 2009 through October 2011, data from 463 consecutive transfused patients were retrospectively reviewed to determine whether Hct correlated with number of PRBC units received. Packed RBC transfusion was not correlated with heart rate and was more highly correlated with Hct (r ¼ 0.45) than with systolic blood pressure or base deficit (r ¼ 0.32 or r ¼ 0.26). Hematocrit was a better overall predictor than systolic blood pressure (sensitivity 45% vs 29%, specificity 94% vs 98%, area under receiver operator characteristic curve 0.71 vs 0.64). Lower Hct was associated with hypotension, more advanced shock, higher blood loss, and increased transfusion of PRBC, plasma, platelets, or cryoprecipitate (all, p < 0.01). Admission Hct is more strongly associated with the PRBC transfusion than either tachycardia, hypotension, or acidosis. Admission Hct is also correlated with 24-hour blood product requirements in those receiving early transfusions. These findings challenge current thinking and suggest that fluid shifts are rapid after trauma and that Hct can be important in initial trauma assessment. (J Am Coll Surg 2013;216:65e73. 2013 by the American College of Surgeons)
relatively slow.1 The axiom that “patients bleed whole blood” has emerged because plasma and red cells are lost in equal amounts.2 In fact, most believe that Hct remains within the normal range immediately after hemorrhage, and that blood loss is not reflected until several hours after a traumatic event.3 However, most physiologists can probably quote references in animal and human research, as well as Starling’s Law of the Capillary, that fluid shifts are rapid due to transcapillary refill after blood loss. We recently reported that admission Hct correlated with hypotension, acidosis, and hemorrhage in patients who required emergency trauma surgery.4 These provocative data challenged conventional wisdom, but the practical application was limited for 2 reasons. First, only those patients who received emergent operations within 4 hours of admission were included. Second, all patients receiving nonoperative management were excluded (eg, angioembolization and those with neurologic and orthopaedic injuries).
Traditionally, hematocrit (Hct) is not considered a reliable index of blood loss after trauma. Classic teaching is that Hct remains stable after a bleeding episode because compensatory mechanisms for fluid shifts are Disclosure Information: Nothing to disclose. Supported in part by grants N140610670 from the Office of Naval Research and W81XWH-11-2-0098 from US Army Medical Research and Material Command. Presented in part at the 59th Annual Florida Chapter American College of Surgeons, Sarasota, FL, May 2012 (First Place Award) and at the American College of Surgeons Florida Committee on Trauma, Miami, FL, November 2011. Received June 12, 2012; Revised September 14, 2012; Accepted September 14, 2012. From the Divisions of Trauma and Surgical Critical Care, Dewitt-Daughtry Family Department of Surgery, University of Miami Miller School of Medicine and Ryder Trauma Center, Miami, FL. Correspondence address: Kenneth G Proctor, PhD, Divisions of Trauma and Surgical Critical Care, Dewitt-Daughtry Family Department of Surgery, University of Miami Miller School of Medicine, Ryder Trauma Center (Suite 215), 1800 NW 10th Ave, Miami, FL 33136. email: [email protected]
ª 2013 by the American College of Surgeons Published by Elsevier Inc.
65
ISSN 1072-7515/12/$36.00 http://dx.doi.org/10.1016/j.jamcollsurg.2012.09.011
Thorson et al
66
Hematocrit after Trauma
Abbreviations and Acronyms
BD Hb Hct HR OR PRBC SBP
¼ ¼ ¼ ¼ ¼ ¼ ¼
base deficit hemoglobin hematocrit heart rate odds ratio packed RBCs systolic blood pressure
This retrospective study examined 2 consecutive samples of trauma patients and was designed to test whether Hct, hypotension, tachycardia, or acidosis correlated with receipt of transfusions and/or the amount transfused during resuscitation. We hypothesized that initial Hct correlates with early transfusion in all trauma patients.
METHODS This was a retrospective review of consecutive patients admitted to a single Level I trauma center within 2 time periods. It was approved by the Institutional Review Board of the University of Miami and the Clinical Trials Office of Jackson Memorial Hospital/ Ryder Trauma Center with waiver of informed consent. Who received transfusions? Adult patients during a 6-month period (June to December 2008) were retrospectively reviewed to identify whether Hct was associated with receipt of blood transfusions. Medical records were examined for those with arrival hemodynamic data and laboratory values. Patients were divided into 2 groups based on transfusion of at least 1 U packed RBCs (PRBC) or no transfusion. How much was transfused? Adults who received PRBCs during the immediate resuscitation period (in the trauma bay and/or operating room) during a 2-year period (October 2009 through October 2011) were retrospectively reviewed to determine whether Hct correlates with transfusion amount. Exclusion criteria were dead on arrival, death during initial resuscitation, pregnant, younger than 18 years old, transferred from outside hospitals, or missing Hct. The patients were divided into the following quartiles: initial Hct <30%, Hct 30% to 34.9%, Hct 35% to 39.9%, and Hct 40%. Demographics and injury information (ie, age, sex, mechanism, Injury Severity Score), vital signs in the resuscitation unit (ie, arrival, minimum/maximum values), 24-hour fluid data (ie, intravenous fluids, blood products, urine output), and laboratory values (ie, blood
J Am Coll Surg
gas pH, base deficit [BD], Hct) were collected. Blood samples were drawn either from the upper extremities or the femoral vein at admission. To ensure standardization of laboratory values, only those run on a blood gas analyzer (Siemens RAPIDLab 1265) in the main laboratory were included. Blood products administered in the first 24 hours were cross checked with blood bank records, and estimated blood loss was obtained from combining blood loss recorded in the operative report with chest tube output. Mortalities were identified from the trauma registry, operative records, and medical examiner reports, when available. Causes of death were assigned by examiners who were unaware if the patient received a transfusion. Statistical analysis PASW software version 19.0 (PASW) was used for statistical analyses. Normally distributed data are reported as mean standard deviation or number (percentage), as appropriate. Data that are not normally distributed (as determined by skewness/kurtosis, Shapiro-Wilk, and Q-Q plots) are expressed as median (interquartile range). Significance was assessed at p < 0.05. Categorical data (ie, sex, mechanism, transfusion status) in Hct subgroups were compared using chi-square test or Fisher’s exact test. Normally distributed data were compared using the Student’s t-test or ANOVA, and nonparametric data were compared with a MannWhitney U test or Kruskal Wallis test. Post-hoc comparisons between Hct groups were done with a Bonferroni correction. Hemodynamic and laboratory variables were used to calculate the sensitivity, specificity, positive predictive value, and negative predictive value for the receipt of a transfusion. To identify ability of vital signs to predict transfusion, receiver operating characteristic curves were generated and areas under the curve were calculated. Variables with a p < 0.20 on univariate analysis were entered into a forward stepwise logistic regression to identify those associated with transfusion. Pearson’s r was used to express the correlation between 2 continuous variables with normal distribution and Spearman’s r was used if the variables were not normally distributed.
RESULTS From June to December 2008, there were 1,492 consecutive admissions with arrival hemodynamic and Hct data. There were 224 (15%) patients who received a transfusion during immediate resuscitation and 1,268 (85%) who did not. Arrival (87 23 beats/min vs 85 22 beats/min) and maximum (118 25 beats/min vs 116 26 beats/min)
Thorson et al
Vol. 216, No. 1, January 2013
heart rates (HR) were comparable between the 2 transfusion groups. In those who received transfusions, systolic blood pressure (SBP) was lower at the scene (108 38 mmHg vs 130 28 mmHg) and on arrival (109 42 mmHg vs 142 26 mmHg) compared with those not transfused (both, p < 0.001). There were more patients in the transfusion group with SBP <90 mmHg alone (n ¼ 66 [30%] vs n ¼ 24 [2%]; p < 0.001), and in combination with a BD 6 mEq/L (n ¼ 33 [15%] vs n ¼ 4 [0.03%]; p < 0.001). Transfused patients had significantly lower Hct on presentation (33% 6% vs 41% 5%; p < 0.001), and were more likely to have Hct <32% (n ¼ 87 [39%] vs n ¼ 43 [3%]; p < 0.001) and < 30 (n ¼ 38 [27%] vs n ¼ 41 [3%]; p < 0.001). Table 1 shows that for assessing receipt of transfusion, there is excellent specificity for SBP at the scene (83% to 96%) and on arrival (91% to 98%), but poor sensitivity (19% to 48%). On the other hand, Hct demonstrated improved sensitivity (29% to 63%) and specificity (87% to 99%). For example, Hct <36% has a sensitivity of 63%, specificity of 87%, and the highest area under the curve (0.770; Table 1). Figure 1 also depicts the higher area under the curve of Hct vs SBP or BD, as denoted by the lines produced on receiver operating characteristic approaching the upper left corner of the graph. Correlations between number of PRBC units transfused and vital signs were higher for Hct (Pearson
Hematocrit after Trauma
67
r ¼ 0.45) than for SBP at the scene (Pearson r ¼ 0.24), SBP on arrival (Pearson r ¼ 0.32), or BD (Spearman r ¼ 0.26). Multiple linear regression revealed a nonlinear relationship between these variables and therefore logistic regression was performed. Hematocrit as a continuous variable (odds ratio [OR] ¼ 0.78; 95% CI, 0.740.80), BD 6 mEq/L (OR ¼ 3.3; 95% CI, 0.210.53), and SBP <90 mmHg (OR ¼ 11.2; 95% CI, 5.921.1) were independently associated with receipt of blood transfusion (overall model r ¼ 0.501; p < 0.001). The OR of 0.78 for Hct represents a 22% increased odds of receiving a transfusion for every unit decrease in Hct. During the 24-month study period, 463 consecutive patients were transfused during initial resuscitation. Mean age was 40 19 years, 76% (n ¼ 350) were male, and 58% (n ¼ 267) sustained blunt injury with an Injury Severity Score of 26 15. Injury mechanisms were most commonly gunshot wounds (n ¼ 137 [30%]), motor vehicle crashes (n ¼ 103 [22%]), and pedestrians struck (n ¼ 91 [20%]). The remainder sustained stab wounds (n ¼ 59 [13%]), motorcycle crashes (n ¼ 53 [11%]), falls (n ¼ 11 [2%]), crush (n ¼ 5 [1%]), or assault (n ¼ 4 [1%]). The disposition from the resuscitation bay was most commonly to the operating room (n ¼ 336 [73%]), with the remainder being admitted to the ICU (n ¼ 111 [24%]) or
Table 1. Sensitivity, Specificity, and Predictive Values of Vital Signs and Laboratory Markers for Receipt of Transfusion in 1,492 Consecutive Trauma Patients from June Through December 2008 Variable
Scene vitals SBP <110 mmHg SBP <90 mmHg Arrival vitals SBP <110 mmHg SBP <90 mmHg BD 6 mEq/L Combination Scene þ arrival SBP < 110 mmHg Scene þ Arrival SBP < 90 mmHg Arrival SBP <110 mmHg and BD 6 mEq/L Arrival SBP <90 mmHg and BD 6 mEq/L Hematocrit <30% <32% <34% <36%
Sensitivity, %
Specificity, %
PPV, %
NPV, %
AUC
43 19
83 96
69 43
89 87
0.632* 0.574*
48 29 37
91 98 90
50 73 40
91 89 89
0.696* 0.638* 0.636*
29 11 25 15
96 99 99 99
58 83 76 89
89 86 88 87
0.629* 0.552y 0.616* 0.572*
29 39 45 63
99 97 94 87
81 67 55 47
89 90 91 93
0.649* 0.692* 0.711* 0.770*
p Values compare the reported area under the curve to the null (0.5). *p < 0.01. y p < 0.05. AUC, area under the curve; BD, base deficit; NPV, negative predictive value; PPV, positive predictive value; SBP, systolic blood pressure.
68
Thorson et al
Hematocrit after Trauma
Figure 1. Receiver operating characteristic curves for (A) conventional vital signs and (B) hematocrit (Hct) with the outcomes of packed RBC transfusion. There was higher sensitivity and specificity of Hct, reflected by higher areas under the curve and line projecting toward the left upper corner, closer to 1.0. BD, base deficit (mEq/L); Hct, hematocrit (%); SBP, systolic blood pressure (mmHg).
hospital ward (n ¼ 16 [3%]). Of those receiving an operation, most (n ¼ 259 [56%]) occurred within the first hour. The mean arrival SBP and HR were 102 40 mmHg and 104 30 beats/min, respectively. A large proportion
J Am Coll Surg
of these patients were hypotensive (n ¼ 173 [37%], SBP <90 mmHg) on arrival. As defined by ATLS guidelines,5 56% (n ¼ 260) arrived in class III/IV shock. At the time of transfusion, 48% (n ¼ 221) were hypotensive and 85% (n ¼ 392) were in class III/IV shock. The median 24-hour blood product requirements were 6 U (interquartile range 8 U) of PRBC and 2 U (interquartile range 6 U) fresh frozen plasma. Table 2 shows demographic data for patients stratified by initial Hct. Those with a lower Hct on arrival were similar with respect to age and mechanism, but there were more females. For vital signs, patients with lower Hct had considerably more acidosis, coagulopathy, and hypotension/shock on arrival and during initial resuscitation. Patients with lower Hct spent less time in the resuscitation bay (61 vs 229 minutes), more received emergent operations (79% vs 62%), and more had a higher mortality (35% vs 8%), comparing Hct <30% with Hct 40% (all, p < 0.05). Overall mortality for the cohort was 22% (n ¼ 100), and of those patients who died, 40% died during initial operative intervention. Exsanguination was responsible for 49% of deaths, 24% were attributed to brain injury, and the remaining 27% occurred due to complications in the ICU (eg, infection, multisystem organ failure, withdrawal of care). Most importantly, there were more deaths due to exsanguination as the admission Hct level decreased (Table 2). Fluid and blood product administration is shown in Table 3, demonstrating considerably higher crystalloid and blood product consumption in patients with lower initial Hct. Those with the lowest Hct (<30%) received a median of 11 U PRBC in the first 24 hours vs groups with higher Hct (8 U vs 6 U vs 4 U), data shown in Figure 2. Likewise, the proportion of patients receiving massive transfusions (>4 U and >10 U) was greater in those with lower initial Hct, and 59% of patients with Hct <30% received >10 U in 24 hours, only 5% of those with Hct 40% received this same amount. The trend of increased blood product requirement was present for fresh frozen plasma, platelets, and cryoprecipitate, shown in Table 3. The estimated blood loss was also greater as Hct decreased (all, p < 0.05). Due to the increased proportion of females in the lower Hct groups, patients were stratified by sex. Because there were few females presenting with Hct 40%, this quartile was combined with the Hct 35% to 39.9% group. The same trends in crystalloid infusion, blood product requirement, and estimated blood loss were present when accounting for the sex imbalance (Table 3).
Thorson et al
Vol. 216, No. 1, January 2013
Hematocrit after Trauma
69
Table 2. Four Hundred Sixty-Three Consecutive Trauma Patients Who Were Transfused from October 2009 Through October 2011 Stratified into Quartiles by Admission Hematocrit Demographics and injury characteristics
Age, mean SD, y Sex, male, n (%) Mechanism, blunt, n (%) Injury Severity Score, mean SD Arrival vital signs SBP, mean SD, mmHg SBP <90 mmHg, n (%) SBP <110 mmHg, n (%) HR, mean SD, beats/min HR >100 beats/min, n (%) Shock class I/II, n (%) Shock class III/IV, n (%) Worst vital signs SBP, mean SD, mmHg SBP <90 mmHg, n (%) HR, mean SD, beats/min HR >100 beats/min, n (%) Arrival laboratory values pH, mean SD BD, median (IQR), mEq/L PT, median (IQR), seconds INR, median (IQR) PTT, median (IQR), seconds Calcium, median (IQR), mmol/L Additional data and outcomes Time in resuscitation, median (IQR), min Disposition, n (%) Operating room ICU Floor Mortality Exsanguination Brain death Late deathz
Hct <30% (n ¼ 97)
Hct 30% to 34.9% (n ¼ 130)
Hct 35% to 39.9% (n ¼ 147)
Hct 40% (n ¼ 89)
p Value
42 21 53 (55) 55 (57) 28 15
41 18 90 (69) 81 (62) 27 16
41 19 125 (85) 70 (54) 25 16
36 15 82 (92) 53 (60) 24 13
0.16 <0.001 0.52 0.26
89 41* 44 (45) 66 (68) 104 37 58 (60) 33 (34) 64 (66)
99 40* 57 (44) 79 (61) 104 30 76 (59) 53 (41) 77 (59)
103 34y 52 (35) 86 (59) 102 27 74 (50) 68 (46) 79 (54)
117 20 35 110 57 47 41
38 (23) (39) 26 (64) (53) (47)
<0.001 0.003 0.001 0.23 0.18 0.05
69 32* 77 (79) 123 31 78 (80)
73 27* 97 (75) 124 27 104 (80)
81 21 100 (68) 121 24 112 (76)
87 27 52 (58) 128 26 75 (84)
<0.001 0.010 0.23 0.51
7.24 0.17* 8 (7)* 13.5 (3.0)* 1.19 (0.26)* 27.5 (8.8)* 1.01 (0.09)*
7.27 0.16 6 (7)* 13.1 (2.2)* 1.15 (0.19)* 25.4 (5.7) 1.03 (0.10)*
7.30 0.14 5 (6) 12.5 (1.5) 1.09 (0.13) 24.4 (4.0) 1.06 (0.08)
7.32 0.11 4 (6) 12.4 (1.9) 1.09 (0.16) 24.5 (4.8) 1.08 (0.10)
<0.001 <0.001 <0.001 <0.001 0.002 0.002
61 (339)*
86 (310)*
123 (574)
229 (662)
0.019 <0.001
77 (79) 20 (21) d 34 (35) 20 (21) 6 (6) 8 (8)
104 25 1 34 18 8 8
(80) (19) (1) (26) (14) (6) (6)
100 41 6 25 10 6 9
(68) (28) (4) (17) (7) (4) (6)
55 25 9 7 1 4 2
(62) (28) (10) (8) (1) (5) (2)
<0.001 <0.001 0.83 0.37
The p values denote a priori treatment effect detected with ANOVA (if parametric) or Kruskal Wallis (if nonparametric). Post-hoc Bonferroni comparisons denoted with hematocrit 40% as reference group. *p < 0.01. y p < 0.05. z Late death is death in ICU due to infection, multisystem organ failure, or withdrawal of care. BD, base deficit; Hct, hematocrit; HR, heart rate; INR, international normalized ratio; IQR, interquartile range; PT, prothrombin time; PTT, partial thromboplastin time.
DISCUSSION The current study showed that Hct was more highly correlated with units of PRBC transfused (r ¼ 0.45) than SBP or BD (r ¼ 0.32 or r ¼ 0.26), and was an independent predictor of transfusion. Patients with lower Hct presented with more hypotension, shock, acidosis, and consumed more fluid and blood products
than those with higher Hct. In addition, admission Hct predicted 24-hour blood product requirements in trauma patients receiving early transfusions. At first glance, it might appear that the practical impact of this study is minimal. The major finding that initial Hct is more strongly correlated with transfusion than low SBP is not surprising. Hypotension after trauma
70
Thorson et al
J Am Coll Surg
Hematocrit after Trauma
Table 3. Twenty-Four-Hour Fluid Requirements in Patients Who Were Transfused from October 2009 Through October 2011 Variable
Total population (n¼463) n Crystalloid, mean SD, mL PRBC, median (IQR), U FFP, median (IQR), U >4 U PRBC, n (%) >10 U PRBC, n (%) Platelets, n (%) Cryoprecipitate, n (%) UOP mean SD, mL EBL, median (IQR), mL Males only (n¼350) n Crystalloid, mean SD, mL PRBC, median (IQR), U FFP, median (IQR), U Platelets, n (%) Cryoprecipitate, n (%) UOP mean SD, mL EBL, median (IQR), mL Females only (n¼113) n Crystalloid, mean SD, mL PRBC, median (IQR), U FFP, median (IQR), U Platelets, n (%) Cryoprecipitate, n (%) UOP mean SD, mL EBL, median (IQR), mL
Hct <30%
Hct 30% to 34.9%
Hct 35% to 39.9%
Hct 40%
p Value
97 9,8185,647* 11 (11)* 5 (9)* 81 (84%) 57 (59%) 51 (53%) 18 (19%) 3,1202,397 1,000 (4,580)*
130 9,3544,512 7 (11)* 4 (8)* 95 (73%) 51 (39%) 50 (39%) 15 (12%) 2,9601,958 1,455 (2,923)*
147 9,4273,904 5 (7)* 0 (5)y 76 (52%) 36 (25%) 38 (26%) 11 (8%) 3,1061,534 750 (2,350)
89 7,7283,004 4 (3) 0 (3) 27 (30%) 4 (5%) 7 (8%) 1 (1%) 2,9731,401 470 (1,290)
0.024 <0.001 <0.001 <0.001 <0.001 <0.001 0.001 0.86 <0.001
53 9,5585,247 11 (10)* 5 (7)* 26 (49) 6 (11) 2,460 (3,210) 1,900 (4,260)*
90 10,1154,896y 8 (14)* 4 (10)* 39 (43) 13 (13) 2,875 (2,156) 1,975 (3,630)*
125 9,9515,796y 6 (7)* 1 (5)y 33 (26) 11 (9) 2,900 (2,020) 1,123 (2,887)y
82 7,8883,932 4 (4) 0 (3) 7 (9) 1 (1) 2,870 (1,511) 525 (1,440)
0.016 <0.001 <0.001 <0.001 0.032 0.88 0.001
40 (4,078)y (5)* (6)* (28) (8) (2,380) (2,380)*
29 (3,621)z (3)z (2)z (17)z d 3,090 (1,645)z 150 (500)z
9,702 11 7 25 12 2,983 675
44 (6,377)* (14)* (12)* (47)* (27) (3,491) (4775)
7,327 5 2 11 3 3,130 1,020
6,041 3 0 5
0.006 <0.001 <0.001 0.001 0.001 0.70 0.004
Data are expressed as Mean SD; Median (IQR); or number (% total observations). p Values denote a priori treatment effect detected with ANOVA (if parametric) or Kruskal Wallis (if nonparametric). Post-hoc Bonferroni comparisons denoted with highest hematocrit group (Hct 40% or Hct 35% in females only) as reference. *p<0.01. y p<0.05. z Hct35%. EBL, estimated blood loss; FFP, fresh frozen plasma; Hct, hematocrit; IQR, interquartile range; PRBC, packed red blood cells; UOP, urine output.
suggests bleeding and the possible need for volume support, and low Hct, in the absence of large-volume fluid therapy, suggests the need for blood. The correlation (r ¼ 0.45) is statistically significant, but would be considered low for prediction. However, the correlations for conventional vital signs were even lower (r ¼ 0.24 to 0.32). Several published guidelines use these conventional vital signs, but not Hct, as transfusion criteria. The decision to administer PRBC can be a lifesaving maneuver, but there are potential consequences for superfluous use, including transfusion reactions, disease transmission, immunosuppression, infections, multisystem organ failure, and transfusion-related acute lung injury.6-8
Numerous evidence-based transfusion guidelines have been developed, but most exclude trauma patients.9-11 The large-scale collaborative, NIH-sponsored “Surgical Glue Grant” attempted to fill this gap.12 A prospective randomized comparison of a “restrictive” hemoglobin (Hb) cutoff (<7 g/dL) vs “liberal” (Hb <10 g/dL) transfusion in the ICU demonstrated lower mortality during hospitalization (22% vs 28%).13 However, subset analysis suggested a more liberal transfusion trigger should be considered in patients with severe cardiovascular disease.14 Transfusion was recommended in hemodynamically stable patients if Hb is <7 g/dL (or Hct <21%). If Hb is >7 g/dL and hypovolemia is present, crystalloid should
Vol. 216, No. 1, January 2013
Figure 2. Box and whisker plots for packed RBC (U) vs admission hematocrit (Hct). From bottom to top, the graph displays gray boxes with the lower quartile (Q1), median (Q2, black line), and upper quartile (Q3). Whiskers extending from the boxes denote sample minimums and maximums. Overall, a trend of increased units of packed RBC transfused is seen with lower presenting hematocrit values.
be infused to achieve normovolemia before considering transfusion.12 Unfortunately, these recommendations are only appropriate for trauma patients after the immediate resuscitation period. A joint task force of the Eastern Association for the Surgery of Trauma and the Society of Critical Care Medicine developed clinical practice guidelines for PRBC transfusion in adult trauma and critical care.15 There is level 1 evidence to support emergent transfusions for patients with hemorrhagic shock, and in those with acute hemorrhage and hemodynamic instability and/or inadequate oxygen delivery. Level 2 evidence suggests that the use of Hb as a transfusion trigger should be avoided, and decisions should be based on intravascular volume status, cardiopulmonary physiology, evidence of shock, and duration and extent of anemia.15 During the primary survey of trauma patients, early and rapid assessment of bleeding and the need for blood volume replacement is crucial. The limitations of conventional vital signs and laboratory values are well established. Systolic blood pressure, for example, might not decrease until 30% to 40% of the total blood volume is lost.5 Although an SBP of <90 mmHg has been the classic cutoff to define hypotension, a recent review of the National Trauma Databank suggested that <110 mmHg might be more clinically relevant.16 Our findings suggest that SBP on arrival has good specificity (91% vs 98% for SBP <110 mmHg vs <90 mmHg), but poor sensitivity (48% vs 29%) for identifying receipt of transfusion.
Thorson et al
Hematocrit after Trauma
71
As a component of the circulation assessment of the trauma patient, it is recommended that blood be drawn for initial Hct, electrolytes, coagulation profile, and blood group typing.17 However, because Hct has a wide range of normal and baseline values for a particular patient are unknown, it is thought to have little value in the initial assessment. Conventional teaching is that Hct remains stable for several hours after trauma because “patients bleed whole blood” and compensatory mechanisms for fluid shifts have not taken place.1-3 There are multiple controlled studies in animals18-22 and humans23-26 showing an almost immediate drop in Hct after hemorrhage with or without fluid resuscitation. In a pressure-driven model of severe hemorrhage in dogs, Prist and colleagues22 analyzed the rate of transcapillary refill using dilution of tagged RBCs. The rate of refill varied directly with the rate of bleeding. The volume of transferred fluid was an ever-increasing fraction of total plasma volume, eventually accounting for >75% of plasma volume in preterminal stages of shock. Kass and colleagues24 found that infusion of 15 mL/kg for 30 minutes instantly decreased the Hct of healthy volunteers by 4.5% 1.3%. Greenfield and colleagues25 found similar results, with a mean Hct drop of 4.5% to 6.4% after 10-, 20- and 30-mL/kg saline boluses. Knottenbelt27 reported reduced Hb in trauma patients with moderate (11.8 g/dL) and severe shock (9.9 g/dL) vs those without shock (12.8 g/dL). Of 31 patients presenting with Hb 8 g/dL, 42% died of hypovolemia compared with only 3% of those with higher levels. In patients with an initial Hct 35%, Snyder28 found a greater need for emergent operation (41% vs 7%) and higher mortality (10% vs 1%). This cutoff was highly specific (90%) for identifying the need for an operation, but lacked sensitivity (50%). Bruns and colleagues29 showed correlations similar to the present study between Hct and SBP (r ¼ 0.301), HR (r ¼ 0.142), and acidosis (r ¼ 0.255). An Hb <10 g/dL in the first 30 minutes of arrival was associated with a 3-fold increase in the need for emergent interventions to stop bleeding.29 Although Hct was not evaluated, Vandromme and colleagues30 demonstrated that blood lactate was a better predictor of transfusion need and mortality than SBP at the scene or in the emergency department. McLaughlin and colleagues31 developed a predictive model for massive transfusion in 302 combat casualties. Those who received massive transfusion had higher mortality (29% vs 7%) and an increased Injury Severity Score (25 11 vs 18 16). Four independent risk factors for massive transfusion were identified: HR >105 beats/min, SBP <110 mmHg, pH <7.25, and Hct <32%.
72
Thorson et al
J Am Coll Surg
Hematocrit after Trauma
Results from the current study confirm what others have already shown; patients requiring a transfusion had substantially lower Hct on arrival (33% vs 41%), initial Hct had improved sensitivity/specificity over conventional vital signs (sensitivity 29% to 63%, specificity 87% to 99%), and had stronger associations with units of PRBC on univariable and multivariable analysis. Altogether, these results are consistent with our previous work4 and suggest that admission Hct can have an important role in the initial assessment of a trauma patient. There have been studies demonstrating poor sensitivity of Hct for identifying bleeding or the need for an emergent operation.28,29,32,33 Therefore, lack of depression in the initial Hct value does not rule out presence of substantial blood loss or ongoing bleeding. In addition, the confounding influence of intravenous fluid and blood product infusion makes interpretation of values difficult,24,25,28 especially if prehospital fluids are unknown (as was the case in the current study). Partly because of these limitations, recent European guidelines do not recommend using a single Hct as an isolated marker for bleeding.34 There are at least 5 potential limitations in this retrospective study. First, Hct was collected based on a standardized clinical protocol rather than a more rigid research protocol. However, laboratory values are obtained on virtually every patient at our trauma center within minutes of arrival; only those who are exceptionally stable with a low suspicion for injury receive no laboratory workup. Second, there was a sex imbalance in the cohort of transfused patients. Stratification by sex was done to address this issue, with similar trends present for all populations investigated. Third, individual variables (ie, Hct, SBP, BD) had relatively poor correlations (r ¼ 0.24 to 0.45) with transfusion, and multiple linear regression could not be performed because of nonlinear relationships between the variables. However, this is not surprising, as there are other factors (ie, baseline anemia, blood pressure medications, intoxication) that can potentially affect resting variables. Fourth, there were 2 separate study populations; 1 included all trauma patients during a 6-month period and the other comprised all trauma patients receiving early transfusion during 24 months. To limit bias and maximize generalizability, both were consecutive samples with limited exclusion criteria. By using 2 distinct, nonoverlapping time periods with 2 databases to address the same question, the chance of sampling bias is reduced and the confidence in our conclusion is strengthened. Lastly, we were unable to measure prehospital fluids, which might have affected
the initial Hct. However, because the mean transport time is <17 minutes and all patients were subject to the same limitation, we believe this does not affect the results substantially.
CONCLUSIONS During the initial resuscitation of a trauma patient, admission Hct was associated with 24-hour fluid and transfusion requirements and had a higher correlation with units of PRBC transfused than SBP, HR, or BD. Although use of the initial Hct might not help clinicians identify all patients who will require a transfusion, it is one more piece of crucial evidence. These results, in context with our previous data, support the idea that fluid shifts are rapid after hemorrhage and that Hct can be valuable during initial trauma assessment. Author Contributions Study conception and design: Thorson, Ryan, Proctor Acquisition of data: Thorson, Van Haren, Ryan, Pereira, Olloqui, Guarch, Barrera, Busko Analysis and interpretation of data: Thorson, Van Haren, Livingstone, Proctor Drafting of manuscript: Thorson, Proctor Critical revision: Thorson, Van Haren, Ryan, Pereira, Olloqui, Guarch, Barrera, Busko, Livingstone, Proctor Acknowledgment: We would like to thank Ronald Manning, RN, BSN, MPH who serves as our research coordinator. REFERENCES 1. Gabrielli A, Layon AJ, Yu M, eds. Civetta, Taylor, & Kirby’s Critical Care. 4th ed. Philadelphia: Lippincott Williams & Wilkins; 2009. 2. Hamilton GC. Emergency Medicine: An Approach to Clinical Problem-Solving. Philadelphia: Saunders; 1991. 3. Gonzalez E, Jastrow K. Hemostasis, surgical bleeding, and transfusion. In: Schwartz SI, Brunicardi FC, eds. Schwartz’s Principles of Surgery. 9th ed. New York: McGraw-Hill, Medical Pub. Division; 2010:xxi. 4. Ryan ML, Thorson CM, Otero CA, et al. Initial hematocrit in trauma: a paradigm shift? J Trauma Acute Care Surg 2012;72: 54e59; discussion 5960. 5. American College of Surgeons Committee on Trauma. Advanced Trauma Life Support (ATLS). 8th ed. Chicago, IL: American College of Surgeons; 2008. 6. Shander A, Popovsky MA. Understanding the consequences of transfusion-related acute lung injury. Chest 2005;128[Suppl 2]:598Se604S. 7. Agarwal N, Murphy JG, Cayten CG, Stahl WM. Blood transfusion increases the risk of infection after trauma. Arch Surg 1993;128:171e176; discussion 176177. 8. Nielsen HJ. Detrimental effects of perioperative blood transfusion. Br J Surg 1995;82:582e587.
Thorson et al
Vol. 216, No. 1, January 2013
9. Practice strategies for elective red blood cell transfusion. American College of Physicians. Ann Intern Med 1992;116: 403e406. 10. Practice guidelines for perioperative blood transfusion and adjuvant therapies: an updated report by the American Society of Anesthesiologists Task Force on Perioperative Blood Transfusion and Adjuvant Therapies. Anesthesiology 2006;105: 198e208. 11. Society of Thoracic Surgeons Blood Conservation Guideline Task Force, Ferraris VA, Ferraris SP, et al. Perioperative blood transfusion and blood conservation in cardiac surgery: the Society of Thoracic Surgeons and The Society of Cardiovascular Anesthesiologists clinical practice guideline. Ann Thorac Surg 2007;83[Suppl]:S27eS86. 12. 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;61:436e439. 13. He´bert 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;340:409e417. Erratum in: N Engl J Med 1999;340:1056. 14. He´bert PC, Yetisir E, Martin C, et al. Is a low transfusion threshold safe in critically ill patients with cardiovascular diseases? Crit Care Med 2001;29:227e234. 15. Napolitano LM, Kurek S, Luchette FA, et al. Clinical practice guideline: red blood cell transfusion in adult trauma and critical care. Crit Care Med 2009;37:3124e3157. Erratum in: Crit Care Med 2010;38:1621. 16. Eastridge BJ, Salinas J, McManus JG, et al. Hypotension begins at 110 mm Hg: redefining “hypotension” with data. J Trauma 2007;63:291e297; discussion 297299. Erratum in: J Trauma 2008;65:501. Concertino, Victor A [corrected to Convertino, Victor A]. 17. Moore FA, Moore EE. Initial management of life-threatening trauma. In: Ashley SW, ed. ACS Surgery: Principles and Practice. Hamilton, ON: Decker Publishing Canada; 2010:1e20. 18. Behrman SW, Fabian TC, Kudsk KA, Proctor KG. Microcirculatory flow changes after initial resuscitation of hemorrhagic shock with 7.5% hypertonic saline/6% dextran 70. J Trauma 1991;31:589e598; discussion 599600. 19. Gibson JB, Maxwell RA, Schweitzer JB, et al. Resuscitation from severe hemorrhagic shock after traumatic brain injury using saline, shed blood, or a blood substitute. Shock 2002; 17:234e244. 20. Jewelewicz DD, Cohn SM, Crookes BA, Proctor KG. Modified rapid deployment hemostat bandage reduces blood loss
21.
22. 23. 24.
25.
26. 27. 28. 29. 30.
31. 32.
33. 34.
Hematocrit after Trauma
73
and mortality in coagulopathic pigs with severe liver injury. J Trauma 2003;55:275e280; discussion 280281. Erratum in J Trauma 2003;55:1e6. Spiers JP, Fabian TC, Kudsk KA, Proctor KG. Resuscitation of hemorrhagic shock with hypertonic saline/dextran or lactated Ringer’s supplemented with AICA riboside. Circ Shock 1993;40:29e36. Prist R, Rocha-e-Silva M, Scalabrini A, et al. A quantitative analysis of transcapillary refill in severe hemorrhagic hypotension in dogs. Shock 1994;1:188e195. Jacobs RG, Howland WS, Goulet AH. Serial microhematocrit determinations in evaluating blood replacement. Anesthesiology 1961;22:342e347. Kass LE, Tien IY, Ushkow BS, Snyder HS. Prospective crossover study of the effect of phlebotomy and intravenous crystalloid on hematocrit. Acad Emerg Med 1997;4: 198e201. Greenfield RH, Bessen HA, Henneman PL. Effect of crystalloid infusion on hematocrit and intravascular volume in healthy, nonbleeding subjects. Ann Emerg Med 1989;18: 51e55. Stamler KD. Effect of crystalloid infusion on hematocrit in nonbleeding patients, with applications to clinical traumatology. Ann Emerg Med 1989;18:747e749. Knottenbelt JD. Low initial hemoglobin levels in trauma patients: an important indicator of ongoing hemorrhage. J Trauma 1991;31:1396e1399. Snyder HS. Significance of the initial spun hematocrit in trauma patients. Am J Emerg Med 1998;16:150e153. Bruns B, Lindsey M, Rowe K, et al. Hemoglobin drops within minutes of injuries and predicts need for an intervention to stop hemorrhage. J Trauma 2007;63:312e315. Vandromme MJ, Griffin RL, Weinberg JA, et al. Lactate is a better predictor than systolic blood pressure for determining blood requirement and mortality: could prehospital measures improve trauma triage? J Am Coll Surg 2010;210:861e867. 867869. McLaughlin DF, Niles SE, Salinas J, et al. A predictive model for massive transfusion in combat casualty patients. J Trauma 2008;64[Suppl]:S57eS63; discussion S63. Opreanu RC, Arrangoiz R, Stevens P, et al. Hematocrit, systolic blood pressure and heart rate are not accurate predictors for surgery to control hemorrhage in injured patients. Am Surg 2010;76:296e301. Paradis NA, Balter S, Davison CM, et al. Hematocrit as a predictor of significant injury after penetrating trauma. Am J Emerg Med 1997;15:224e228. Rossaint R, Bouillon B, Cerny V, et al. Management of bleeding following major trauma: an updated European guideline. Crit Care 2010;14:R52.
METHODS:
RESULTS:
CONCLUSION:
BS,
The decision to transfuse packed RBCs (PRBC) during initial resuscitation of trauma patients is based on physiologic state, evidence for blood loss, and potential for ongoing hemorrhage. Initial hematocrit (Hct) is not considered an accurate marker of blood loss. This study tests the hypothesis that admission Hct is associated with transfusion requirements after trauma. From June to December 2008, data from 1,492 consecutive admissions at a Level I trauma center were retrospectively reviewed to determine whether initial Hct was associated with PRBC transfusions. From October 2009 through October 2011, data from 463 consecutive transfused patients were retrospectively reviewed to determine whether Hct correlated with number of PRBC units received. Packed RBC transfusion was not correlated with heart rate and was more highly correlated with Hct (r ¼ 0.45) than with systolic blood pressure or base deficit (r ¼ 0.32 or r ¼ 0.26). Hematocrit was a better overall predictor than systolic blood pressure (sensitivity 45% vs 29%, specificity 94% vs 98%, area under receiver operator characteristic curve 0.71 vs 0.64). Lower Hct was associated with hypotension, more advanced shock, higher blood loss, and increased transfusion of PRBC, plasma, platelets, or cryoprecipitate (all, p < 0.01). Admission Hct is more strongly associated with the PRBC transfusion than either tachycardia, hypotension, or acidosis. Admission Hct is also correlated with 24-hour blood product requirements in those receiving early transfusions. These findings challenge current thinking and suggest that fluid shifts are rapid after trauma and that Hct can be important in initial trauma assessment. (J Am Coll Surg 2013;216:65e73. 2013 by the American College of Surgeons)
relatively slow.1 The axiom that “patients bleed whole blood” has emerged because plasma and red cells are lost in equal amounts.2 In fact, most believe that Hct remains within the normal range immediately after hemorrhage, and that blood loss is not reflected until several hours after a traumatic event.3 However, most physiologists can probably quote references in animal and human research, as well as Starling’s Law of the Capillary, that fluid shifts are rapid due to transcapillary refill after blood loss. We recently reported that admission Hct correlated with hypotension, acidosis, and hemorrhage in patients who required emergency trauma surgery.4 These provocative data challenged conventional wisdom, but the practical application was limited for 2 reasons. First, only those patients who received emergent operations within 4 hours of admission were included. Second, all patients receiving nonoperative management were excluded (eg, angioembolization and those with neurologic and orthopaedic injuries).
Traditionally, hematocrit (Hct) is not considered a reliable index of blood loss after trauma. Classic teaching is that Hct remains stable after a bleeding episode because compensatory mechanisms for fluid shifts are Disclosure Information: Nothing to disclose. Supported in part by grants N140610670 from the Office of Naval Research and W81XWH-11-2-0098 from US Army Medical Research and Material Command. Presented in part at the 59th Annual Florida Chapter American College of Surgeons, Sarasota, FL, May 2012 (First Place Award) and at the American College of Surgeons Florida Committee on Trauma, Miami, FL, November 2011. Received June 12, 2012; Revised September 14, 2012; Accepted September 14, 2012. From the Divisions of Trauma and Surgical Critical Care, Dewitt-Daughtry Family Department of Surgery, University of Miami Miller School of Medicine and Ryder Trauma Center, Miami, FL. Correspondence address: Kenneth G Proctor, PhD, Divisions of Trauma and Surgical Critical Care, Dewitt-Daughtry Family Department of Surgery, University of Miami Miller School of Medicine, Ryder Trauma Center (Suite 215), 1800 NW 10th Ave, Miami, FL 33136. email: [email protected]
ª 2013 by the American College of Surgeons Published by Elsevier Inc.
65
ISSN 1072-7515/12/$36.00 http://dx.doi.org/10.1016/j.jamcollsurg.2012.09.011
Thorson et al
66
Hematocrit after Trauma
Abbreviations and Acronyms
BD Hb Hct HR OR PRBC SBP
¼ ¼ ¼ ¼ ¼ ¼ ¼
base deficit hemoglobin hematocrit heart rate odds ratio packed RBCs systolic blood pressure
This retrospective study examined 2 consecutive samples of trauma patients and was designed to test whether Hct, hypotension, tachycardia, or acidosis correlated with receipt of transfusions and/or the amount transfused during resuscitation. We hypothesized that initial Hct correlates with early transfusion in all trauma patients.
METHODS This was a retrospective review of consecutive patients admitted to a single Level I trauma center within 2 time periods. It was approved by the Institutional Review Board of the University of Miami and the Clinical Trials Office of Jackson Memorial Hospital/ Ryder Trauma Center with waiver of informed consent. Who received transfusions? Adult patients during a 6-month period (June to December 2008) were retrospectively reviewed to identify whether Hct was associated with receipt of blood transfusions. Medical records were examined for those with arrival hemodynamic data and laboratory values. Patients were divided into 2 groups based on transfusion of at least 1 U packed RBCs (PRBC) or no transfusion. How much was transfused? Adults who received PRBCs during the immediate resuscitation period (in the trauma bay and/or operating room) during a 2-year period (October 2009 through October 2011) were retrospectively reviewed to determine whether Hct correlates with transfusion amount. Exclusion criteria were dead on arrival, death during initial resuscitation, pregnant, younger than 18 years old, transferred from outside hospitals, or missing Hct. The patients were divided into the following quartiles: initial Hct <30%, Hct 30% to 34.9%, Hct 35% to 39.9%, and Hct 40%. Demographics and injury information (ie, age, sex, mechanism, Injury Severity Score), vital signs in the resuscitation unit (ie, arrival, minimum/maximum values), 24-hour fluid data (ie, intravenous fluids, blood products, urine output), and laboratory values (ie, blood
J Am Coll Surg
gas pH, base deficit [BD], Hct) were collected. Blood samples were drawn either from the upper extremities or the femoral vein at admission. To ensure standardization of laboratory values, only those run on a blood gas analyzer (Siemens RAPIDLab 1265) in the main laboratory were included. Blood products administered in the first 24 hours were cross checked with blood bank records, and estimated blood loss was obtained from combining blood loss recorded in the operative report with chest tube output. Mortalities were identified from the trauma registry, operative records, and medical examiner reports, when available. Causes of death were assigned by examiners who were unaware if the patient received a transfusion. Statistical analysis PASW software version 19.0 (PASW) was used for statistical analyses. Normally distributed data are reported as mean standard deviation or number (percentage), as appropriate. Data that are not normally distributed (as determined by skewness/kurtosis, Shapiro-Wilk, and Q-Q plots) are expressed as median (interquartile range). Significance was assessed at p < 0.05. Categorical data (ie, sex, mechanism, transfusion status) in Hct subgroups were compared using chi-square test or Fisher’s exact test. Normally distributed data were compared using the Student’s t-test or ANOVA, and nonparametric data were compared with a MannWhitney U test or Kruskal Wallis test. Post-hoc comparisons between Hct groups were done with a Bonferroni correction. Hemodynamic and laboratory variables were used to calculate the sensitivity, specificity, positive predictive value, and negative predictive value for the receipt of a transfusion. To identify ability of vital signs to predict transfusion, receiver operating characteristic curves were generated and areas under the curve were calculated. Variables with a p < 0.20 on univariate analysis were entered into a forward stepwise logistic regression to identify those associated with transfusion. Pearson’s r was used to express the correlation between 2 continuous variables with normal distribution and Spearman’s r was used if the variables were not normally distributed.
RESULTS From June to December 2008, there were 1,492 consecutive admissions with arrival hemodynamic and Hct data. There were 224 (15%) patients who received a transfusion during immediate resuscitation and 1,268 (85%) who did not. Arrival (87 23 beats/min vs 85 22 beats/min) and maximum (118 25 beats/min vs 116 26 beats/min)
Thorson et al
Vol. 216, No. 1, January 2013
heart rates (HR) were comparable between the 2 transfusion groups. In those who received transfusions, systolic blood pressure (SBP) was lower at the scene (108 38 mmHg vs 130 28 mmHg) and on arrival (109 42 mmHg vs 142 26 mmHg) compared with those not transfused (both, p < 0.001). There were more patients in the transfusion group with SBP <90 mmHg alone (n ¼ 66 [30%] vs n ¼ 24 [2%]; p < 0.001), and in combination with a BD 6 mEq/L (n ¼ 33 [15%] vs n ¼ 4 [0.03%]; p < 0.001). Transfused patients had significantly lower Hct on presentation (33% 6% vs 41% 5%; p < 0.001), and were more likely to have Hct <32% (n ¼ 87 [39%] vs n ¼ 43 [3%]; p < 0.001) and < 30 (n ¼ 38 [27%] vs n ¼ 41 [3%]; p < 0.001). Table 1 shows that for assessing receipt of transfusion, there is excellent specificity for SBP at the scene (83% to 96%) and on arrival (91% to 98%), but poor sensitivity (19% to 48%). On the other hand, Hct demonstrated improved sensitivity (29% to 63%) and specificity (87% to 99%). For example, Hct <36% has a sensitivity of 63%, specificity of 87%, and the highest area under the curve (0.770; Table 1). Figure 1 also depicts the higher area under the curve of Hct vs SBP or BD, as denoted by the lines produced on receiver operating characteristic approaching the upper left corner of the graph. Correlations between number of PRBC units transfused and vital signs were higher for Hct (Pearson
Hematocrit after Trauma
67
r ¼ 0.45) than for SBP at the scene (Pearson r ¼ 0.24), SBP on arrival (Pearson r ¼ 0.32), or BD (Spearman r ¼ 0.26). Multiple linear regression revealed a nonlinear relationship between these variables and therefore logistic regression was performed. Hematocrit as a continuous variable (odds ratio [OR] ¼ 0.78; 95% CI, 0.740.80), BD 6 mEq/L (OR ¼ 3.3; 95% CI, 0.210.53), and SBP <90 mmHg (OR ¼ 11.2; 95% CI, 5.921.1) were independently associated with receipt of blood transfusion (overall model r ¼ 0.501; p < 0.001). The OR of 0.78 for Hct represents a 22% increased odds of receiving a transfusion for every unit decrease in Hct. During the 24-month study period, 463 consecutive patients were transfused during initial resuscitation. Mean age was 40 19 years, 76% (n ¼ 350) were male, and 58% (n ¼ 267) sustained blunt injury with an Injury Severity Score of 26 15. Injury mechanisms were most commonly gunshot wounds (n ¼ 137 [30%]), motor vehicle crashes (n ¼ 103 [22%]), and pedestrians struck (n ¼ 91 [20%]). The remainder sustained stab wounds (n ¼ 59 [13%]), motorcycle crashes (n ¼ 53 [11%]), falls (n ¼ 11 [2%]), crush (n ¼ 5 [1%]), or assault (n ¼ 4 [1%]). The disposition from the resuscitation bay was most commonly to the operating room (n ¼ 336 [73%]), with the remainder being admitted to the ICU (n ¼ 111 [24%]) or
Table 1. Sensitivity, Specificity, and Predictive Values of Vital Signs and Laboratory Markers for Receipt of Transfusion in 1,492 Consecutive Trauma Patients from June Through December 2008 Variable
Scene vitals SBP <110 mmHg SBP <90 mmHg Arrival vitals SBP <110 mmHg SBP <90 mmHg BD 6 mEq/L Combination Scene þ arrival SBP < 110 mmHg Scene þ Arrival SBP < 90 mmHg Arrival SBP <110 mmHg and BD 6 mEq/L Arrival SBP <90 mmHg and BD 6 mEq/L Hematocrit <30% <32% <34% <36%
Sensitivity, %
Specificity, %
PPV, %
NPV, %
AUC
43 19
83 96
69 43
89 87
0.632* 0.574*
48 29 37
91 98 90
50 73 40
91 89 89
0.696* 0.638* 0.636*
29 11 25 15
96 99 99 99
58 83 76 89
89 86 88 87
0.629* 0.552y 0.616* 0.572*
29 39 45 63
99 97 94 87
81 67 55 47
89 90 91 93
0.649* 0.692* 0.711* 0.770*
p Values compare the reported area under the curve to the null (0.5). *p < 0.01. y p < 0.05. AUC, area under the curve; BD, base deficit; NPV, negative predictive value; PPV, positive predictive value; SBP, systolic blood pressure.
68
Thorson et al
Hematocrit after Trauma
Figure 1. Receiver operating characteristic curves for (A) conventional vital signs and (B) hematocrit (Hct) with the outcomes of packed RBC transfusion. There was higher sensitivity and specificity of Hct, reflected by higher areas under the curve and line projecting toward the left upper corner, closer to 1.0. BD, base deficit (mEq/L); Hct, hematocrit (%); SBP, systolic blood pressure (mmHg).
hospital ward (n ¼ 16 [3%]). Of those receiving an operation, most (n ¼ 259 [56%]) occurred within the first hour. The mean arrival SBP and HR were 102 40 mmHg and 104 30 beats/min, respectively. A large proportion
J Am Coll Surg
of these patients were hypotensive (n ¼ 173 [37%], SBP <90 mmHg) on arrival. As defined by ATLS guidelines,5 56% (n ¼ 260) arrived in class III/IV shock. At the time of transfusion, 48% (n ¼ 221) were hypotensive and 85% (n ¼ 392) were in class III/IV shock. The median 24-hour blood product requirements were 6 U (interquartile range 8 U) of PRBC and 2 U (interquartile range 6 U) fresh frozen plasma. Table 2 shows demographic data for patients stratified by initial Hct. Those with a lower Hct on arrival were similar with respect to age and mechanism, but there were more females. For vital signs, patients with lower Hct had considerably more acidosis, coagulopathy, and hypotension/shock on arrival and during initial resuscitation. Patients with lower Hct spent less time in the resuscitation bay (61 vs 229 minutes), more received emergent operations (79% vs 62%), and more had a higher mortality (35% vs 8%), comparing Hct <30% with Hct 40% (all, p < 0.05). Overall mortality for the cohort was 22% (n ¼ 100), and of those patients who died, 40% died during initial operative intervention. Exsanguination was responsible for 49% of deaths, 24% were attributed to brain injury, and the remaining 27% occurred due to complications in the ICU (eg, infection, multisystem organ failure, withdrawal of care). Most importantly, there were more deaths due to exsanguination as the admission Hct level decreased (Table 2). Fluid and blood product administration is shown in Table 3, demonstrating considerably higher crystalloid and blood product consumption in patients with lower initial Hct. Those with the lowest Hct (<30%) received a median of 11 U PRBC in the first 24 hours vs groups with higher Hct (8 U vs 6 U vs 4 U), data shown in Figure 2. Likewise, the proportion of patients receiving massive transfusions (>4 U and >10 U) was greater in those with lower initial Hct, and 59% of patients with Hct <30% received >10 U in 24 hours, only 5% of those with Hct 40% received this same amount. The trend of increased blood product requirement was present for fresh frozen plasma, platelets, and cryoprecipitate, shown in Table 3. The estimated blood loss was also greater as Hct decreased (all, p < 0.05). Due to the increased proportion of females in the lower Hct groups, patients were stratified by sex. Because there were few females presenting with Hct 40%, this quartile was combined with the Hct 35% to 39.9% group. The same trends in crystalloid infusion, blood product requirement, and estimated blood loss were present when accounting for the sex imbalance (Table 3).
Thorson et al
Vol. 216, No. 1, January 2013
Hematocrit after Trauma
69
Table 2. Four Hundred Sixty-Three Consecutive Trauma Patients Who Were Transfused from October 2009 Through October 2011 Stratified into Quartiles by Admission Hematocrit Demographics and injury characteristics
Age, mean SD, y Sex, male, n (%) Mechanism, blunt, n (%) Injury Severity Score, mean SD Arrival vital signs SBP, mean SD, mmHg SBP <90 mmHg, n (%) SBP <110 mmHg, n (%) HR, mean SD, beats/min HR >100 beats/min, n (%) Shock class I/II, n (%) Shock class III/IV, n (%) Worst vital signs SBP, mean SD, mmHg SBP <90 mmHg, n (%) HR, mean SD, beats/min HR >100 beats/min, n (%) Arrival laboratory values pH, mean SD BD, median (IQR), mEq/L PT, median (IQR), seconds INR, median (IQR) PTT, median (IQR), seconds Calcium, median (IQR), mmol/L Additional data and outcomes Time in resuscitation, median (IQR), min Disposition, n (%) Operating room ICU Floor Mortality Exsanguination Brain death Late deathz
Hct <30% (n ¼ 97)
Hct 30% to 34.9% (n ¼ 130)
Hct 35% to 39.9% (n ¼ 147)
Hct 40% (n ¼ 89)
p Value
42 21 53 (55) 55 (57) 28 15
41 18 90 (69) 81 (62) 27 16
41 19 125 (85) 70 (54) 25 16
36 15 82 (92) 53 (60) 24 13
0.16 <0.001 0.52 0.26
89 41* 44 (45) 66 (68) 104 37 58 (60) 33 (34) 64 (66)
99 40* 57 (44) 79 (61) 104 30 76 (59) 53 (41) 77 (59)
103 34y 52 (35) 86 (59) 102 27 74 (50) 68 (46) 79 (54)
117 20 35 110 57 47 41
38 (23) (39) 26 (64) (53) (47)
<0.001 0.003 0.001 0.23 0.18 0.05
69 32* 77 (79) 123 31 78 (80)
73 27* 97 (75) 124 27 104 (80)
81 21 100 (68) 121 24 112 (76)
87 27 52 (58) 128 26 75 (84)
<0.001 0.010 0.23 0.51
7.24 0.17* 8 (7)* 13.5 (3.0)* 1.19 (0.26)* 27.5 (8.8)* 1.01 (0.09)*
7.27 0.16 6 (7)* 13.1 (2.2)* 1.15 (0.19)* 25.4 (5.7) 1.03 (0.10)*
7.30 0.14 5 (6) 12.5 (1.5) 1.09 (0.13) 24.4 (4.0) 1.06 (0.08)
7.32 0.11 4 (6) 12.4 (1.9) 1.09 (0.16) 24.5 (4.8) 1.08 (0.10)
<0.001 <0.001 <0.001 <0.001 0.002 0.002
61 (339)*
86 (310)*
123 (574)
229 (662)
0.019 <0.001
77 (79) 20 (21) d 34 (35) 20 (21) 6 (6) 8 (8)
104 25 1 34 18 8 8
(80) (19) (1) (26) (14) (6) (6)
100 41 6 25 10 6 9
(68) (28) (4) (17) (7) (4) (6)
55 25 9 7 1 4 2
(62) (28) (10) (8) (1) (5) (2)
<0.001 <0.001 0.83 0.37
The p values denote a priori treatment effect detected with ANOVA (if parametric) or Kruskal Wallis (if nonparametric). Post-hoc Bonferroni comparisons denoted with hematocrit 40% as reference group. *p < 0.01. y p < 0.05. z Late death is death in ICU due to infection, multisystem organ failure, or withdrawal of care. BD, base deficit; Hct, hematocrit; HR, heart rate; INR, international normalized ratio; IQR, interquartile range; PT, prothrombin time; PTT, partial thromboplastin time.
DISCUSSION The current study showed that Hct was more highly correlated with units of PRBC transfused (r ¼ 0.45) than SBP or BD (r ¼ 0.32 or r ¼ 0.26), and was an independent predictor of transfusion. Patients with lower Hct presented with more hypotension, shock, acidosis, and consumed more fluid and blood products
than those with higher Hct. In addition, admission Hct predicted 24-hour blood product requirements in trauma patients receiving early transfusions. At first glance, it might appear that the practical impact of this study is minimal. The major finding that initial Hct is more strongly correlated with transfusion than low SBP is not surprising. Hypotension after trauma
70
Thorson et al
J Am Coll Surg
Hematocrit after Trauma
Table 3. Twenty-Four-Hour Fluid Requirements in Patients Who Were Transfused from October 2009 Through October 2011 Variable
Total population (n¼463) n Crystalloid, mean SD, mL PRBC, median (IQR), U FFP, median (IQR), U >4 U PRBC, n (%) >10 U PRBC, n (%) Platelets, n (%) Cryoprecipitate, n (%) UOP mean SD, mL EBL, median (IQR), mL Males only (n¼350) n Crystalloid, mean SD, mL PRBC, median (IQR), U FFP, median (IQR), U Platelets, n (%) Cryoprecipitate, n (%) UOP mean SD, mL EBL, median (IQR), mL Females only (n¼113) n Crystalloid, mean SD, mL PRBC, median (IQR), U FFP, median (IQR), U Platelets, n (%) Cryoprecipitate, n (%) UOP mean SD, mL EBL, median (IQR), mL
Hct <30%
Hct 30% to 34.9%
Hct 35% to 39.9%
Hct 40%
p Value
97 9,8185,647* 11 (11)* 5 (9)* 81 (84%) 57 (59%) 51 (53%) 18 (19%) 3,1202,397 1,000 (4,580)*
130 9,3544,512 7 (11)* 4 (8)* 95 (73%) 51 (39%) 50 (39%) 15 (12%) 2,9601,958 1,455 (2,923)*
147 9,4273,904 5 (7)* 0 (5)y 76 (52%) 36 (25%) 38 (26%) 11 (8%) 3,1061,534 750 (2,350)
89 7,7283,004 4 (3) 0 (3) 27 (30%) 4 (5%) 7 (8%) 1 (1%) 2,9731,401 470 (1,290)
0.024 <0.001 <0.001 <0.001 <0.001 <0.001 0.001 0.86 <0.001
53 9,5585,247 11 (10)* 5 (7)* 26 (49) 6 (11) 2,460 (3,210) 1,900 (4,260)*
90 10,1154,896y 8 (14)* 4 (10)* 39 (43) 13 (13) 2,875 (2,156) 1,975 (3,630)*
125 9,9515,796y 6 (7)* 1 (5)y 33 (26) 11 (9) 2,900 (2,020) 1,123 (2,887)y
82 7,8883,932 4 (4) 0 (3) 7 (9) 1 (1) 2,870 (1,511) 525 (1,440)
0.016 <0.001 <0.001 <0.001 0.032 0.88 0.001
40 (4,078)y (5)* (6)* (28) (8) (2,380) (2,380)*
29 (3,621)z (3)z (2)z (17)z d 3,090 (1,645)z 150 (500)z
9,702 11 7 25 12 2,983 675
44 (6,377)* (14)* (12)* (47)* (27) (3,491) (4775)
7,327 5 2 11 3 3,130 1,020
6,041 3 0 5
0.006 <0.001 <0.001 0.001 0.001 0.70 0.004
Data are expressed as Mean SD; Median (IQR); or number (% total observations). p Values denote a priori treatment effect detected with ANOVA (if parametric) or Kruskal Wallis (if nonparametric). Post-hoc Bonferroni comparisons denoted with highest hematocrit group (Hct 40% or Hct 35% in females only) as reference. *p<0.01. y p<0.05. z Hct35%. EBL, estimated blood loss; FFP, fresh frozen plasma; Hct, hematocrit; IQR, interquartile range; PRBC, packed red blood cells; UOP, urine output.
suggests bleeding and the possible need for volume support, and low Hct, in the absence of large-volume fluid therapy, suggests the need for blood. The correlation (r ¼ 0.45) is statistically significant, but would be considered low for prediction. However, the correlations for conventional vital signs were even lower (r ¼ 0.24 to 0.32). Several published guidelines use these conventional vital signs, but not Hct, as transfusion criteria. The decision to administer PRBC can be a lifesaving maneuver, but there are potential consequences for superfluous use, including transfusion reactions, disease transmission, immunosuppression, infections, multisystem organ failure, and transfusion-related acute lung injury.6-8
Numerous evidence-based transfusion guidelines have been developed, but most exclude trauma patients.9-11 The large-scale collaborative, NIH-sponsored “Surgical Glue Grant” attempted to fill this gap.12 A prospective randomized comparison of a “restrictive” hemoglobin (Hb) cutoff (<7 g/dL) vs “liberal” (Hb <10 g/dL) transfusion in the ICU demonstrated lower mortality during hospitalization (22% vs 28%).13 However, subset analysis suggested a more liberal transfusion trigger should be considered in patients with severe cardiovascular disease.14 Transfusion was recommended in hemodynamically stable patients if Hb is <7 g/dL (or Hct <21%). If Hb is >7 g/dL and hypovolemia is present, crystalloid should
Vol. 216, No. 1, January 2013
Figure 2. Box and whisker plots for packed RBC (U) vs admission hematocrit (Hct). From bottom to top, the graph displays gray boxes with the lower quartile (Q1), median (Q2, black line), and upper quartile (Q3). Whiskers extending from the boxes denote sample minimums and maximums. Overall, a trend of increased units of packed RBC transfused is seen with lower presenting hematocrit values.
be infused to achieve normovolemia before considering transfusion.12 Unfortunately, these recommendations are only appropriate for trauma patients after the immediate resuscitation period. A joint task force of the Eastern Association for the Surgery of Trauma and the Society of Critical Care Medicine developed clinical practice guidelines for PRBC transfusion in adult trauma and critical care.15 There is level 1 evidence to support emergent transfusions for patients with hemorrhagic shock, and in those with acute hemorrhage and hemodynamic instability and/or inadequate oxygen delivery. Level 2 evidence suggests that the use of Hb as a transfusion trigger should be avoided, and decisions should be based on intravascular volume status, cardiopulmonary physiology, evidence of shock, and duration and extent of anemia.15 During the primary survey of trauma patients, early and rapid assessment of bleeding and the need for blood volume replacement is crucial. The limitations of conventional vital signs and laboratory values are well established. Systolic blood pressure, for example, might not decrease until 30% to 40% of the total blood volume is lost.5 Although an SBP of <90 mmHg has been the classic cutoff to define hypotension, a recent review of the National Trauma Databank suggested that <110 mmHg might be more clinically relevant.16 Our findings suggest that SBP on arrival has good specificity (91% vs 98% for SBP <110 mmHg vs <90 mmHg), but poor sensitivity (48% vs 29%) for identifying receipt of transfusion.
Thorson et al
Hematocrit after Trauma
71
As a component of the circulation assessment of the trauma patient, it is recommended that blood be drawn for initial Hct, electrolytes, coagulation profile, and blood group typing.17 However, because Hct has a wide range of normal and baseline values for a particular patient are unknown, it is thought to have little value in the initial assessment. Conventional teaching is that Hct remains stable for several hours after trauma because “patients bleed whole blood” and compensatory mechanisms for fluid shifts have not taken place.1-3 There are multiple controlled studies in animals18-22 and humans23-26 showing an almost immediate drop in Hct after hemorrhage with or without fluid resuscitation. In a pressure-driven model of severe hemorrhage in dogs, Prist and colleagues22 analyzed the rate of transcapillary refill using dilution of tagged RBCs. The rate of refill varied directly with the rate of bleeding. The volume of transferred fluid was an ever-increasing fraction of total plasma volume, eventually accounting for >75% of plasma volume in preterminal stages of shock. Kass and colleagues24 found that infusion of 15 mL/kg for 30 minutes instantly decreased the Hct of healthy volunteers by 4.5% 1.3%. Greenfield and colleagues25 found similar results, with a mean Hct drop of 4.5% to 6.4% after 10-, 20- and 30-mL/kg saline boluses. Knottenbelt27 reported reduced Hb in trauma patients with moderate (11.8 g/dL) and severe shock (9.9 g/dL) vs those without shock (12.8 g/dL). Of 31 patients presenting with Hb 8 g/dL, 42% died of hypovolemia compared with only 3% of those with higher levels. In patients with an initial Hct 35%, Snyder28 found a greater need for emergent operation (41% vs 7%) and higher mortality (10% vs 1%). This cutoff was highly specific (90%) for identifying the need for an operation, but lacked sensitivity (50%). Bruns and colleagues29 showed correlations similar to the present study between Hct and SBP (r ¼ 0.301), HR (r ¼ 0.142), and acidosis (r ¼ 0.255). An Hb <10 g/dL in the first 30 minutes of arrival was associated with a 3-fold increase in the need for emergent interventions to stop bleeding.29 Although Hct was not evaluated, Vandromme and colleagues30 demonstrated that blood lactate was a better predictor of transfusion need and mortality than SBP at the scene or in the emergency department. McLaughlin and colleagues31 developed a predictive model for massive transfusion in 302 combat casualties. Those who received massive transfusion had higher mortality (29% vs 7%) and an increased Injury Severity Score (25 11 vs 18 16). Four independent risk factors for massive transfusion were identified: HR >105 beats/min, SBP <110 mmHg, pH <7.25, and Hct <32%.
72
Thorson et al
J Am Coll Surg
Hematocrit after Trauma
Results from the current study confirm what others have already shown; patients requiring a transfusion had substantially lower Hct on arrival (33% vs 41%), initial Hct had improved sensitivity/specificity over conventional vital signs (sensitivity 29% to 63%, specificity 87% to 99%), and had stronger associations with units of PRBC on univariable and multivariable analysis. Altogether, these results are consistent with our previous work4 and suggest that admission Hct can have an important role in the initial assessment of a trauma patient. There have been studies demonstrating poor sensitivity of Hct for identifying bleeding or the need for an emergent operation.28,29,32,33 Therefore, lack of depression in the initial Hct value does not rule out presence of substantial blood loss or ongoing bleeding. In addition, the confounding influence of intravenous fluid and blood product infusion makes interpretation of values difficult,24,25,28 especially if prehospital fluids are unknown (as was the case in the current study). Partly because of these limitations, recent European guidelines do not recommend using a single Hct as an isolated marker for bleeding.34 There are at least 5 potential limitations in this retrospective study. First, Hct was collected based on a standardized clinical protocol rather than a more rigid research protocol. However, laboratory values are obtained on virtually every patient at our trauma center within minutes of arrival; only those who are exceptionally stable with a low suspicion for injury receive no laboratory workup. Second, there was a sex imbalance in the cohort of transfused patients. Stratification by sex was done to address this issue, with similar trends present for all populations investigated. Third, individual variables (ie, Hct, SBP, BD) had relatively poor correlations (r ¼ 0.24 to 0.45) with transfusion, and multiple linear regression could not be performed because of nonlinear relationships between the variables. However, this is not surprising, as there are other factors (ie, baseline anemia, blood pressure medications, intoxication) that can potentially affect resting variables. Fourth, there were 2 separate study populations; 1 included all trauma patients during a 6-month period and the other comprised all trauma patients receiving early transfusion during 24 months. To limit bias and maximize generalizability, both were consecutive samples with limited exclusion criteria. By using 2 distinct, nonoverlapping time periods with 2 databases to address the same question, the chance of sampling bias is reduced and the confidence in our conclusion is strengthened. Lastly, we were unable to measure prehospital fluids, which might have affected
the initial Hct. However, because the mean transport time is <17 minutes and all patients were subject to the same limitation, we believe this does not affect the results substantially.
CONCLUSIONS During the initial resuscitation of a trauma patient, admission Hct was associated with 24-hour fluid and transfusion requirements and had a higher correlation with units of PRBC transfused than SBP, HR, or BD. Although use of the initial Hct might not help clinicians identify all patients who will require a transfusion, it is one more piece of crucial evidence. These results, in context with our previous data, support the idea that fluid shifts are rapid after hemorrhage and that Hct can be valuable during initial trauma assessment. Author Contributions Study conception and design: Thorson, Ryan, Proctor Acquisition of data: Thorson, Van Haren, Ryan, Pereira, Olloqui, Guarch, Barrera, Busko Analysis and interpretation of data: Thorson, Van Haren, Livingstone, Proctor Drafting of manuscript: Thorson, Proctor Critical revision: Thorson, Van Haren, Ryan, Pereira, Olloqui, Guarch, Barrera, Busko, Livingstone, Proctor Acknowledgment: We would like to thank Ronald Manning, RN, BSN, MPH who serves as our research coordinator. REFERENCES 1. Gabrielli A, Layon AJ, Yu M, eds. Civetta, Taylor, & Kirby’s Critical Care. 4th ed. Philadelphia: Lippincott Williams & Wilkins; 2009. 2. Hamilton GC. Emergency Medicine: An Approach to Clinical Problem-Solving. Philadelphia: Saunders; 1991. 3. Gonzalez E, Jastrow K. Hemostasis, surgical bleeding, and transfusion. In: Schwartz SI, Brunicardi FC, eds. Schwartz’s Principles of Surgery. 9th ed. New York: McGraw-Hill, Medical Pub. Division; 2010:xxi. 4. Ryan ML, Thorson CM, Otero CA, et al. Initial hematocrit in trauma: a paradigm shift? J Trauma Acute Care Surg 2012;72: 54e59; discussion 5960. 5. American College of Surgeons Committee on Trauma. Advanced Trauma Life Support (ATLS). 8th ed. Chicago, IL: American College of Surgeons; 2008. 6. Shander A, Popovsky MA. Understanding the consequences of transfusion-related acute lung injury. Chest 2005;128[Suppl 2]:598Se604S. 7. Agarwal N, Murphy JG, Cayten CG, Stahl WM. Blood transfusion increases the risk of infection after trauma. Arch Surg 1993;128:171e176; discussion 176177. 8. Nielsen HJ. Detrimental effects of perioperative blood transfusion. Br J Surg 1995;82:582e587.
Thorson et al
Vol. 216, No. 1, January 2013
9. Practice strategies for elective red blood cell transfusion. American College of Physicians. Ann Intern Med 1992;116: 403e406. 10. Practice guidelines for perioperative blood transfusion and adjuvant therapies: an updated report by the American Society of Anesthesiologists Task Force on Perioperative Blood Transfusion and Adjuvant Therapies. Anesthesiology 2006;105: 198e208. 11. Society of Thoracic Surgeons Blood Conservation Guideline Task Force, Ferraris VA, Ferraris SP, et al. Perioperative blood transfusion and blood conservation in cardiac surgery: the Society of Thoracic Surgeons and The Society of Cardiovascular Anesthesiologists clinical practice guideline. Ann Thorac Surg 2007;83[Suppl]:S27eS86. 12. 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;61:436e439. 13. He´bert 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;340:409e417. Erratum in: N Engl J Med 1999;340:1056. 14. He´bert PC, Yetisir E, Martin C, et al. Is a low transfusion threshold safe in critically ill patients with cardiovascular diseases? Crit Care Med 2001;29:227e234. 15. Napolitano LM, Kurek S, Luchette FA, et al. Clinical practice guideline: red blood cell transfusion in adult trauma and critical care. Crit Care Med 2009;37:3124e3157. Erratum in: Crit Care Med 2010;38:1621. 16. Eastridge BJ, Salinas J, McManus JG, et al. Hypotension begins at 110 mm Hg: redefining “hypotension” with data. J Trauma 2007;63:291e297; discussion 297299. Erratum in: J Trauma 2008;65:501. Concertino, Victor A [corrected to Convertino, Victor A]. 17. Moore FA, Moore EE. Initial management of life-threatening trauma. In: Ashley SW, ed. ACS Surgery: Principles and Practice. Hamilton, ON: Decker Publishing Canada; 2010:1e20. 18. Behrman SW, Fabian TC, Kudsk KA, Proctor KG. Microcirculatory flow changes after initial resuscitation of hemorrhagic shock with 7.5% hypertonic saline/6% dextran 70. J Trauma 1991;31:589e598; discussion 599600. 19. Gibson JB, Maxwell RA, Schweitzer JB, et al. Resuscitation from severe hemorrhagic shock after traumatic brain injury using saline, shed blood, or a blood substitute. Shock 2002; 17:234e244. 20. Jewelewicz DD, Cohn SM, Crookes BA, Proctor KG. Modified rapid deployment hemostat bandage reduces blood loss
21.
22. 23. 24.
25.
26. 27. 28. 29. 30.
31. 32.
33. 34.
Hematocrit after Trauma
73
and mortality in coagulopathic pigs with severe liver injury. J Trauma 2003;55:275e280; discussion 280281. Erratum in J Trauma 2003;55:1e6. Spiers JP, Fabian TC, Kudsk KA, Proctor KG. Resuscitation of hemorrhagic shock with hypertonic saline/dextran or lactated Ringer’s supplemented with AICA riboside. Circ Shock 1993;40:29e36. Prist R, Rocha-e-Silva M, Scalabrini A, et al. A quantitative analysis of transcapillary refill in severe hemorrhagic hypotension in dogs. Shock 1994;1:188e195. Jacobs RG, Howland WS, Goulet AH. Serial microhematocrit determinations in evaluating blood replacement. Anesthesiology 1961;22:342e347. Kass LE, Tien IY, Ushkow BS, Snyder HS. Prospective crossover study of the effect of phlebotomy and intravenous crystalloid on hematocrit. Acad Emerg Med 1997;4: 198e201. Greenfield RH, Bessen HA, Henneman PL. Effect of crystalloid infusion on hematocrit and intravascular volume in healthy, nonbleeding subjects. Ann Emerg Med 1989;18: 51e55. Stamler KD. Effect of crystalloid infusion on hematocrit in nonbleeding patients, with applications to clinical traumatology. Ann Emerg Med 1989;18:747e749. Knottenbelt JD. Low initial hemoglobin levels in trauma patients: an important indicator of ongoing hemorrhage. J Trauma 1991;31:1396e1399. Snyder HS. Significance of the initial spun hematocrit in trauma patients. Am J Emerg Med 1998;16:150e153. Bruns B, Lindsey M, Rowe K, et al. Hemoglobin drops within minutes of injuries and predicts need for an intervention to stop hemorrhage. J Trauma 2007;63:312e315. Vandromme MJ, Griffin RL, Weinberg JA, et al. Lactate is a better predictor than systolic blood pressure for determining blood requirement and mortality: could prehospital measures improve trauma triage? J Am Coll Surg 2010;210:861e867. 867869. McLaughlin DF, Niles SE, Salinas J, et al. A predictive model for massive transfusion in combat casualty patients. J Trauma 2008;64[Suppl]:S57eS63; discussion S63. Opreanu RC, Arrangoiz R, Stevens P, et al. Hematocrit, systolic blood pressure and heart rate are not accurate predictors for surgery to control hemorrhage in injured patients. Am Surg 2010;76:296e301. Paradis NA, Balter S, Davison CM, et al. Hematocrit as a predictor of significant injury after penetrating trauma. Am J Emerg Med 1997;15:224e228. Rossaint R, Bouillon B, Cerny V, et al. Management of bleeding following major trauma: an updated European guideline. Crit Care 2010;14:R52.