clinical investigations in critical care

Transcription

1 clinical investigations in critical care RBC Transfusion in the ICU* Is There a Reason? Howard L. Corwin, MD, FCCP; Kathy C. Parsonnet, RN, MPH; and Andrew G ettinger, MD, FCCP Objective: To evaluate RBC transfusion practice in the ICU. Design: Retrospective chart review. Setting: Multidisciplinary ICU in a tertiary care center. Patients: All patients admitted to the ICU with a length of stay of greater than 1 week. Results: A total of 23% of all patients admitted to the ICU had a length of stay of greater than 1 week (19.6:±: 1.6 days). Of these patients, 85% received blood transfusions (9.5:±:0.8 U per patient). These transfusions were not solely a function of acute blood loss. Patients were transfused a constant 2 to 3 U/wk. Patients receiving blood transfusions were phlebotomized on average 61 to 70 ml per day. Phlebotomy accounted for 49% of the variation in amount of RBCs transfused. No indication for blood transfusion was identified for 29% of transfusion events. A low hemat- ocrit ( <25%) was the only identifiable indication in an additional19% of events. Almost one third of all RBCs transfused were without a clear transfusion indication. Conclusion: The long-term ICU population receive a large number of blood transfusions. Phlebotomy contributes significantly to these transfusions. There is no clear indication for a large number of the blood transfusions given. Many blood transfusions appear to be administered because of an arbitrary "transfusion trigger" rather than a physiologic need for blood. Blood conservation and adherence to transfusion guidelines could significantly reduce RBC transfusion in the ICU. (CHEST 1995; 108:767-71) Key words: blood transfusion; ICU anemia; transfusion guidelines; transfusion indication; transfusion trigger Transfusion practice is under intense evaluation. 1-3 This is in part the result of a heightened s ensitivity of both the medical and lay communities to transfusion-related infection, in particular HIV. Infection is not the only potential adverse effect of blood transfusion nor HN the only potentially infecting agent. 4 5 Some adverse effect to the recipient may occur in as many as 20% of blood transfusions. 5 Although most of these adverse e vents are minor in nature, they may still result in morbidity. At a time when transfusion practice, in general, is under scrutiny, httle data are available regarding transfusion behavior in the critical care setting. We undertook a study of the long-term ICU patient population (> l week length of stay) to determine the magnitude of their blood transfusion requirement, the factors related to the need for a blood transfusion, and the specific indications for blood transfusion. We chose *From the Critical Care Service and the Departments o f Anesthesiology and Medicine, Dartmouth-Hitchcock Medical Center, Lebanon, NH. Supported b y Ortho Biotech, Raritan, NT. Dr. Corwin has received support from the Elsie 0. and Philip 0. Sang Foundation. Manuscript received S eptember 26, 1994; revision accepted F ebruary 1, f995. Reprint requests: Dr. Corwin, Dartmouth-Hitchcock M edical Center, One Medical Center Drive, Lebanon, OH the long-term ICU patient population for study because blood use in these patients has more potential to be affected by interventions altering either transfusion practice or need. General M ETHODS The ICU at Dartmouth-Hitchcock Medical Center is a multidisciplinary unit. It is an open unit; however, all patients are followed up by the Critical Care Service. A blood transfusion can be ordered b y either the primary care team or the ICU team. At the time of the study, there was no transfusion practice protocol in place. All patients admitted t o the multidisciplinary ICU at Dartmouth Hitchcock Medical Center during 1990, with an ICU length of stay greater than l week, were included in the study. D ata were obtained from institutional databases and by retrospective chart review. The amount of blood phlebotomized f or each ptient a was estimated from the number of diagnostic tests performed and the milliliters of blood required for each t est. Blood Transfusion Events A transfusion event was considered to have occurred if any RBCs were transfused in a 24-h p eriod (12 midnight to 12 midnight). All blood transfused in a 24-h period was considered t o be a sult re of the single transfusion event. This was based on the assumption that, in general, all blood transfused during the 24-h period constituting a transfusion event would be for the same indication. Only one transfusion event was counted f or a given 24-h. CHEST I 108 I 3 I SEPTEMBER,

2 Transfusion Indication An indication for each transfusion event was based on preestablished criteria that follows at end of this paragraph and determined by retrospective review of the hospital record. The indication was not required to be documented in the hospital chart, but rather was determined retrospectively based on the clinical and laboratory data available in the chart. Each transfusion event was evaluated independently and considered unrelated to prior transfusion events for a given patient. Thus, one patient could have several transfusion events, each for a different indication. However, each event could have only one indication. If more than one indication applied to an event, it was considered to be multifactorial. The specific transfusion indications were adapted from the National Institutes of Health Consensus Conference transfusion criteria. Indications included active bleeding or surgery, low hematocrit ( <25% ), low cardiac output, myocardial infarction/ischemia, oxygen transport, renal ischemia, and preoperative. If none of the above indications were present, the transfusion event was classified as having no indication. Statistics The relationships between variables were examined using either the Student's t test, analysis of variance (Fisher least significant difference for multiple comparisons), x 2, or multiple linear regression, where appropriate. Significance was considered to be p <0.05. Data are presented as SEM. Patients RESULTS A total of 609 patients were admitted to the ICU during the study period. One hundred forty-two (23%) had an ICU stay of greater than 1 week (mean, 19.6:±: 1.6 days) and were included in the study. Of these 142 patients, 121 (85%) were transfused while in the ICU. Patient characteristics are displayed in Table 1. Transfusions The 121 patients who received at least one RBC transfusion were transfused a total of 1,147 U of blood (9.5:±:0.8 U per patient). Of the 121 patients whoreceived blood, 49 ( 40%) were transfused only during the first week of their ICU admission. However, patients (72) transfused after week 1 had a constant transfusion requirement of 2 to 3 U per patient per week No single variable or combination of variables at the time of ICU admission predicted those patients who were transfused after the initial week in the ICU. Low vs High Transfusion We compared patients who received no blood transfusion with those receiving either >5 U (50th percentile) of blood (high transfusion) or 1 to 5 U of blood (low transfusion). The low transfusion group received 2.9:±:0.2 U and the high group received 13.9:±:1.1 U (p<0.0001). The comparisons among the three groups are displayed in Table 1. The hematocrit on admission to the ICU was significantly lower in patients with a higher transfusion requirement. The high transfusion group required more prolonged mechani- 768 Table!-Transfusion Requirement: High vs Low* No. of patients Units transfused Age, yr Sex (female/male) Service (med!surg or trauma) APACHE day 1 ICU, d Hospital, d Ventilator, d ICU admission hematocrit ICU survival, % Hospital survival, % *Significant differences (p<0.05). 1 None vs >5: 1-5 vs >5. 'None vs 1-5: None vs >5.!None vs 1-5: None vs >5; and 1-5 vs >5. Amount of Blood Transfused, U None ±4 9/12 20/1 18±1.5 15±4 28±4 13±2 39± > ± ± ± 2 62± 2 26/26 29/40 37/15 43/26' 21±1 20±1 14± 1 25±31 36±3 41±4 12±1 19± ±1 30± cal ventilation, had a longer ICU stay, and a higher mortality. Similar results (not shown) were found if 10 U (75th percentile) was used to stratify the group receiving blood transfusion into a low and a high group (4.6:±:2.6 and 20.2:±:1.2 U, respectively; p<o.oool). Phlebotomy The 142 patients studied were phlebotomized a total of 176 L of blood for diagnostic tests during their ICU stay. Patients who did not require blood transfusion were phlebotomized significantly less than those who were transfused (Table 2). Patients in the high transfusion group (either >5 U or >10 U) had a daily phlebotomy blood loss comparable to the patients in the low transfusion group (1 to 5 U). The higher total phlebotomy blood loss in the patients in the higher transfusion group reflected longer ICU stays (Table 1). Phlebotomy blood loss, similar to blood transfusion, remained at a constant level throughout the ICU course. The 176 L of blood phlebotomized (assuming an average hematocrit of 30%) account for about 300 U of RBCs or 30% of the total RBCs transfused. Based on multiple regression analysis, 49% of the variation in Transfusions None 1-5 u >5 u >10 u *Significant differences (p<0.05). Table 2-Phlebotomy* 601±77 858±59 1,708±137 2,156±208 Blood Drawn, ml Daily! 40±5 61±4 68±6 70±6 1 None vs 1-5; None vs >5; None vs >10; 1-5 vs >5; and 1-5 vs >10. INone vs 1-5; None vs >5; and None vs >10. Clinical Investigations in Critical Care

3 Table 3--Transfusion Indication Total Indication % Pre-HCT* Nonacute,% None Surgery/bleeding Low hematocrit Low cardiac output Myocardial ische mia Oxygen transport Other *HCT =hematocrit. the amount of red blood cells transfused is accounted for by phlebotomy (p<o.ool). Transfusion Events and Indications The 121 patients who were transfused in the ICU received a total of 1,147 U of blood during 478 separate transfusion events. The indications for the transfusion events are displayed in Table 3. No indication could be identified for 29% of transfusion events and a low hematocrit was the only identifiable indication in a further 19% of transfusion events. The pre transfusion hematocrit was about 27% for all indications, with the exception of those transfusions performed for low hematocrit (set at 25%). These latter events tended to occur significantly later during the ICU course. Acute vs Nonacute Blood Loss There were 165 (35%) transfusion events associated with acute blood loss (surgery or active bleeding within 24 h of transfusion event). Of the 313 (65%) transfusion events for nonacute blood loss, 63% were performed for either no identifiable indication or for low hematocrit alone (Table 3). The total nonacute blood transfused was 579 U or 50% of the total blood transfusion requirement for these patients in the ICU. Those transfusions for either no identified indication or low hematocrit accounted for 320 U or 28% of the total RBCs transfused. DISCUSSION We have shown that our patients in the ICU long term receive an extremely high number of RBC transfusions. During the period of this study, these patients comprised 1.5% of the total patients admitted to our hospital and accounted for 4% of total inpatient days; however, they received 16% of all inpatient RBC transfusions. These transfusions were not simply a response to acute blood loss, rather they were persistent throughout a patient's ICU course. Phlebotomy was a major factor contributing to RBC transfusion. The equivalent of 30% of the total blood transfused was phlebotomized from our patients during their ICU stay. This is similar to a report by Smoller and Kruskall 6 who noted that almost half of their patients in the ICU receiving blood transfusions were phlebotomized more than the equivalent of 1 U of blood. The patients in the ICU described in their study on average were phlebotomized 65 mud, also similar to our findings (Table 2). Phlebotomy blood losses in this range are consistent with other reports of patients in the ICU and are associated with the development of anemia. 7-9 The importance of blood conservation in the critical care setting has been stressed by Chemow. 10 Given the strong association of phlebotomy with blood transfusion in the ICU, reduction of this daily blood loss should impact positively on blood use in the ICU. Approaches directed toward reducing phlebotomy blood loss include use of small-volume (pediatric) tubes, elimination of arterial line blood discard, elimination of standing orders for laboratory tests, altering test ordering behavior, and daily feedback The use of small-volume (pediatric) tubes has been shown to result in a decrease in phlebotomy blood loss of between 33 to 47%. 7 8 Foulke and Harlow 7 noted a significant reduction in the num her of patients transfused coinciding with a reduction in the daily phlebotomy blood loss. This is consistent with our finding that patients who were not transfused had significantly lower daily phlebotomy blood loss. Recently two blood conservation systems designed to eliminate blood discards have been reported. 12.l 3 Such systems in combination with advances in in-line monitoring of laboratory parameters, ie, arterial blood gases, could greatly reduce phlebotomy blood loss. We did not specifically look at the appropriateness of the laboratory tests ordered, but this clearly is an important issue. Why do we transfuse when we do? The variability in transfusion practice between individual physicians and institutions is striking Salem-Schatz et ap 9 have reported widespread deficiencies in physicians' knowledge of transfusion risks and indications. Our finding that almost half of all transfusion events were performed for either no identifiable indication or low hematocrit alone is consistent with a lack of an understanding blood transfusion indication. Other studies looking at transfusion indications have reported the number of inqppropriate transfusions to range from 4% to as high as 57% In this study, there was no specific transfusion protocol in place in our ICU. Our data suggest that transfusion decisions in our patients may have been driven by a specific "transfusion trigger," a hematocrit of27%. We found that the pretransfusion hematocrit was at this level regardless of transfusion indication (Table 3). Although retrospective review did identify a possible transfusion indication at the time of many of the transfusion events, the lack of any significant variation in pretransfusion hematocrit suggests to us that it was CHEST /108 I 31 SEPTEMBER,

4 the "trigger" rather than the indication that actually drove the decision to transfuse. The "transfusion trigger" may in part explain why some patients were not transfused. The nontransfused patients had admission hematocrits significantly above that seen in transfused patients and well above transfusion trigger in our ICU. These patients were therefore less likely to have their hematocrit fall with phlebotomy, other blood loss, or both to a level that would trigger a blood transfusion. Historically a hemoglobin level of 10 gldl and hematocrit of 30% have been the minimum levels believed to be adequate, particularly in the surgical setting. 23 Over recent years, data have accumulated suggesting that much lower levels of hemoglobin are tolerated by patients without adverse effects. In patients undergoing surgery who refused blood transfusion, hemoglobin levels as low as 7 to 8 gldl are physiologically well tolerated. 24 Kitchen 25 has summarized the recent published experience in Jehovah's Witness patients undergoing major surgery and noted only a 1.4% mortality attributable to anemia. Similarly, studies in both dogs and baboons show that hematocrits in the 10 to 20% range can be maintained without untoward effects to the animals In patients in the ICU, recent studies have not shown physiologic benefit to augmenting oxygen delivery via blood transfusion in patients with sepsis and shock What the optimal hematocrit is for the ICU patient remains to be determined. How much blood could potentially be saved in these patients? Current transfusion guidelines stress the importance of abandoning specific "transfusion triggers" in making transfusion decisions, in particular in patients with hemoglobin levels above 7 gldl. 1-3 In our patients, transfusions performed at a time when there was no clinical evidence of acute blood loss accounted for 50% of the total blood transfusion requirement in these patients in the ICU. Of these events, 63% were either without identifiable indication or for low hematocrit (Table 3). These events account for 28% of the total blood transfused and could potentially be saved. The real savings could be higher if some transfusions for "clinical indications" actually reflect a "transfusion trigger" effect. Transfusions associated with acute. blood loss are less likely to be eliminated, however, to the extent that these transfusion decisions, driven by arbitrary "triggers," could be decreased. The combination of altering transfusion practice with decreasing phlebotomy blood loss could act synergistically to decrease the amount of blood transfused in the ICU. This would particularly benefit patients in the low transfusion group (1 to 5 U), who if made transfusion free would be spared all transfusion-associated morbidity. In summary, we have shown that the long-term ICU patient population receives a large number of RBC no transfusions. A large proportion of these transfusions appear to be a response to an arbitrary "transfusion trigger" rather than a response to a specific clinical or physiologic indication for blood. Phlebotomy blood loss is a major factor contributing to RBC transfusion in the ICU. Decreasing phlebotomy blood loss and altering transfusion practice to correspond to recent guidelines 1 2 will result in a significant reduction in blood use in the ICU. REFERENCES 1 Consensus Conference. Perioperative red blood cell transfusion. JAMA 1988; 260: Practice strategies for elective blood cell transfusion. Ann Intern Med 1992; 116: Welch HG, Meehan KR, Goodnough LT. Prudent strategies for elective red blood cell transfusion. Ann Intern Med 1992; 116: Tartter Pl. Blood transfusion and postoperative infections. Transfusion 1989; 29: Walker RH. Transfusion risks. Am J Clin Pathol1987; 88: Smoller BR, Kruskall MS. Phlebotomy for diagnostic laboratory tests in adults: pattern of use and effect on transfusion requirements. N Eng! J Med 1986; 314: Foulke GE, Harlow OJ. Effective measures for reducing blood loss from diagnostic laboratory tests in intensive care unit patients. Crit Care Med 1989; 17: Smoller BR, Kruskall MS, Horowitz GL. Reducing adult phlebotomy blood loss with the use of pediatric sized blood collection tubes. Am J Clin Pathol1989; 91: Eyster E, Bernene J. Nosocomial anemia. JAMA 1973; 223:73-4 lo Chernow B. Blood conservation in critical care--the evidence accumulates. Crit Care Med 1993; 21: Civetta JM, Hudson-Civetta JA. Maintaining quality of care while reducing charges in the ICU. Ann Surg 1985; 202: Peruzzi WT, Parker MA, Lichtenthal PR, et a!. A clinical evaluation of a blood conservation device in medical intensive care unit patients. Crit Care Med 1993; 21: Silver MJ, Jubran H, Stein S, eta!. Evaluation of a n ew blood conserving arterial line system for intensive care units. Crit Care Med 1993; 21:507-ll 14 Zimmerman JL, Dellinger RP. Initial evaluation of a new intraarterial blood gas system in humans. Crit Care Med 1993; 21: Shapiro BA, Mahutte CK, Cane RD, eta!. Clinical performance of a blood gas monitor: a prospective multicenter trial. Crit Care Med 1993; 21: Hardy JF, Tremblay N, Robitaille D, eta!. The stratification of cardiac surgical procedures according to use of blood products: a retrospective analysis of 1480 cases. Can J Anaesth 1991; 38: Katz AJ, Palermo G. Patterns of blood use in Connecticut. Transfusion 1980; 20: Goodnough LT, Johnston MFM, Toy PTCY. The variability of transfusion practice in coronary artery bypass surgery. JAMA 1991; 265: Salem-Schatz SR, Avom J, Soumerai SB. Influence of clinical knowledge, organizational context, and practice style on transfusion decision making. JAMA 1991; 264: Brien WF, Butler RJ, Inwood MJ. An audit of blood component therapy in a Canadian general teaching hospital. Can Med Assoc J 1989; 140: Coffin C, Matz K, Rich E. Algorithms for evaluating the appropriateness of blood transfusion. Transfusion 1989; 29: Clinical lnvesbgations in Crttical Care

5 22 Mozes B, Epstein M, Ben-Bassat I, et al. Evaluation of the appropriateness of blood and blood product transfusion using preset criteria. Transfusion 1989; 29: Cane RD. Hemoglobin: how much is enough? Crit Care Med 1990; 18: Spence RK, Carson JA, Poses R, et al. Elective surgery without transfusion: influence of preoperative hemoglobin level and blood loss on mortality. Am J Surg 1990; 159: Kitchen CS. Are transfusions overrated? surgical outcome of Jehovah's Witnesses. Am J M ed 1993; 94: Levine E, Rosen A, Sehgal L, et a!. Physiologic effects of acute anemia: implications for a reduced transfusion trigger. Transfusion 1990; 30: Geha AS. Coronary and cardiovascular dynamics and oxygen availability during acute normovolemic anemia. Surgery 1976; 80: Rosen AL, Gould SA, Sehgal LR, et a!. Erythropoietic response to acute anemia. Crit Care Med 1990; 18: Dietrich KA, Conrad SA, Hebert CA, et a!. Cardiovascular and metabolic response to red blood transfusion in critically ill volume resuscitated patients. Crit Care Med 1990; 18: Mink RB, Pollack MM. Effect of blood transfusion on oxygen consumption in pediatric septic shock. Crit Care Med 1990; 18: More Study Opportunities in Critical Care Medicine CHEST {1995} October 29 - November 2, 1995 New York, New York CHEST 1995* offers you thought-provoking sessions on current issues in chest medicine, presented by world-renowned experts--the same quality science found each month in Chest. If this article on critical care medicine was beneficial, you will want to take advantage of the educational opportunities at CHEST 1995-where critical care medicine will be one of four study tracks offered. For more information, call ACCP Product and Registration Services: or fonm:rly Annuallntmuztional Scimti.fic Asstmb/y CHEST / 108/3/ SEPTEMBER, 1995 n1