Red Blood Cell Transfusion Practices: A College of American Pathologists Q-Probes Study of Compliance With Audit Criteria in 128 Hospitals

Participating hospitals in the College of American Pathologists 2008 Q-Probes study 83, “Utilization of RBC Transfusions,” retrospectively reviewed a target of 50 recent, consecutive, nonexcluded RBC-transfusion episodes for compliance with their institutional guidelines.²  An episode was defined as one or more RBC units issued at the same time. For each episode, the pretransfusion and posttransfusion hemoglobin (Hgb) or hematocrit (Hct) levels, the category of indication for transfusion, and the applicable institutional Hgb/Hct threshold for compliance review were provided. Institutions that use posttransfusion Hgb/Hct levels for auditing intraoperative transfusions provided those values. Hematocrits were multiplied by 0.34 for expression as Hgb.

Transfusions were excluded from the study if they were autologous or given to pediatric or trauma patients.

For the most-recent calendar year, participants provided their numbers of RBC transfusions performed and their percentages of RBC units reviewed for, and in compliance with, institutional criteria. They provided their RBC transfusion guidelines for common indications for RBC transfusions, including whether adjustments were made for underlying cardiovascular or pulmonary disease.

Demographic features of participating institutions are shown in Table 1. Seventy-four percent (78 of 106) of the participant hospitals had 300 beds or fewer. One-third (32%, 35 of 109) were teaching hospitals, and 19% (21 of 112) had a pathology residency program.

During the year before the Q-Probes study, 131 participating facilities (94% [123 of 131] in the United States) transfused 724 332 RBC units, for a mean of 5659 units per facility. Those facilities transfused approximately 4.5% of all the RBCs given in the United States at the time of the study.³  The participants' median annual transfusions for other components were 296 plateletpheresis units, 0 whole-blood–derived platelets, 972 plasma units, and 109 cryoprecipitate units. Eighty-eight percent (114 of 130) of the participants transfused autologous RBCs, mostly in small proportions. In 51% (66 of 130) of facilities, autologous units were 1% or less of the RBCs used, and in 15% (20 of 130), they comprised 1.1% to 3.0% of the RBCs transfused. Eighty-three percent (108 of 130) had a blood salvage program.

During this Q-Probes study, 128 participating institutions provided complete data on 5830 transfusion events, for an average of 45.6 events per facility, near the goal of 50 events per participant. Most of the reviewed events involved 1 (60%; n = 3924) or 2 (32%; n = 2063) units. The times of issue were midnight to 5:59 am, 11.8% (n = 764); 6:00 am to noon, 29.8% (n = 1922); 12:01 pm to 6:00 pm, 37.7% (n = 2433); and 6:01 pm to 11:59 pm, 20.7% (n = 1332).

The median pretransfusion and posttransfusion Hgb results for all individual transfusions combined were 8.1 g/dL (range, 2.3–16.5 g/dL) (Table 2) and 9.8 g/dL (range, 3.8–17.9 g/dL), respectively. Each institution's average pretransfusion Hgb value from among all transfusions in the study was calculated. Fewer than 10% (11 of 128) of the participants had an average pretransfusion Hgb above 8.8 g/dL.

The 4 most-common categories of transfusion settings, accounting for 80% (5179 of 6102) of the episodes combined, were nonhemorrhagic anemia, 36.5% (n = 2372); nonoperative hemorrhage, 20.5% (n = 1334); postoperative hemorrhage, 13.8% (n = 894), and intraoperative hemorrhage, 8.9% (n = 579). Ten percent (n = 644) were counted under “no indication provided by clinician.”

The blood-ordering process was available via computer in 69% (89 of 129) of facilities and/or via paper in 54% (70 of 130). The facilities provided lists of clinical indications for blood products in 23% (20 of 89) of systems with computerized order entry and in 43% (30 of 70) of systems using paper orders. Overall, the institutional transfusion guidelines were readily available to ordering physicians in 73% (94 of 129) of the facilities. Educational sessions on transfusion practice had been held in 47% (60 of 128) of the facilities in the previous year.

When RBCs were ordered, 39% (49 of 127) of the participants required submission of a clinical indication for transfusion in all cases, 18% (23 of 127) did in some cases, and 43% (55 of 127) did not require an indication.

Participants reported their sources for transfusion guidelines as follows (multiple answers permitted): journal article(s), 70% (n = 88); professional organization, 63% (n = 79); textbook(s), 62% (n = 78); local expert opinion, 45% (n = 57); local blood donor center, 18% (n = 22); centralized requirements of a health care system, 14% (n = 17); and unknown, 6% (n = 8). Sixty-six percent (85 of 129) had modified their criteria in the previous 5 years in response to publications on risks and benefits of transfusion.

The institutional criteria for the pretransfusion Hgb level that would trigger a review were dependent on whether cardiovascular or pulmonary disease was present. Without the latter conditions, the median review thresholds were in the range of 8.0 to 8.9 g/dL Hgb across all of the following patient categories: preoperative, intraoperative with bleeding, postoperative with bleeding, critical illness, hematology/oncology, and cases that were nonsurgical-nonhematology-nonbleeding. If any of those types of patients had cardiovascular or pulmonary disease, the median review thresholds were 9.0 to 9.9 g/dL, with the following 2 exceptions: the median threshold was 10.0 g/dL in nonsurgical-nonhematology-nonbleeding patients with cardiovascular/pulmonary disease, and the question on critical illness did not include pulmonary disease, only cardiovascular disease.

Table 3 shows the percentage of participants that employed various review elements. Notably, only 9% (12 of 129) audited one-unit transfusions, and only 16% had different criteria for autologous transfusions.

During the year before the Q-Probes study, the participants, in aggregate, had reviewed usage in 37.4% (270 900 of 724 332) of all RBC units transfused. The median percentage of transfusions reviewed among all participants was 50% (range, 0%–100%) (Table 2). Forty-one percent (54 of 131) of participant facilities reviewed all RBC transfusions.

Higher percentages of transfusions reviewed in the year before the Q-Probes study were associated with low platelet and cryoprecipitate usage, absence of a teaching program, absence of a local expert opinion for guidelines, and the use of contact notices to individual transfusing physicians about noncompliant transfusions (Table 4).

During the Q-Probes study period, 63.6% (3709 of 5830) of all transfusion events were compliant with facility guidelines. The median percentage of institutionally compliant transfusions across all participants was 69.0% (range, 0–100) (Table 2). During the year before the Q-Probes study, 93.1% (252 208 of 270 900) of transfusions had been found to be in compliance (median retrospective compliance rate, 99.1%, Table 2).

The rates of institutional compliance during the Q-Probes study were similar to the overall average across most transfusion indications. Transfusions for anemia, preoperative anemia, hemorrhage, postoperative hemorrhage, and hemolytic anemia all met criteria in 63% to 68% of cases. However, transfusions for intraoperative hemorrhage met criteria in only 37% (187 of 503) of cases (P < .001). Correspondingly, institutions with more than 10% of their transfusions for intraoperative hemorrhage had a lower median percentage of meeting guidelines (61.2%), compared with those with 0% to 5% transfusions for intraoperative hemorrhage (79.3%). Thirty-one percent of facilities (39 of 126) used postoperative Hgb/Hct results to assess intraoperative transfusions (Table 3).

Table 5 shows communication policies for notifications about noncompliant transfusions. Thirty-five percent of participating facilities required compliance with transfusion guidelines for physician practice privileges.

One-third (33.6%; 42 of 125) of the participants had a review threshold of at least 10 g/dL Hgb pretransfusion values for patients without cardiovascular or pulmonary disease. Those facilities were less likely to have a pathologist or transfusion medicine specialist actively involved in transfusion practice consultations (45%; 19 of 42), versus those who had more strict review criteria (64% [53 of 83] with active involvement).

One-third (33 of 98) of reporting institutions did not have a hematology/oncology physician on their transfusion committees. Those institutions had a lower median percentage of transfusion compliance during the Q-Probes study compared with those who had a hematologist/oncologist on the committee (57% versus 79% compliance rates).

The College of American Pathologists Q-Probes studies yield valuable insights into patient care, laboratory practices, and quality improvement benchmarking by providing forums for participating institutions to conveniently collect and share focused data during short periods on carefully chosen topics. Q-Probes and their siblings Q-Tracks, which provide longitudinal quality improvement trending on fundamental measures of laboratory quality, are widely cited resources in laboratory medicine.^4–6  In transfusion medicine, previous studies have addressed numerous aspects of quality practice, including specimen-labeling accuracy,⁷  preoperative type-and-screen testing,⁸  RBC crossmatching,⁹  transfusion administration,¹⁰  and blood component wastage.^9,11 

This Q-Probes study examined use-review practices for RBC transfusions. One hundred twenty-eight institutions with nearly three-quarter of a million aggregate, annual RBC transfusions participated in the study. Facilities with less complexity—that is, nonteaching institutions, institutions with no transfusion expert for their guidelines, and institutions performing fewer platelet and cryoprecipitate transfusions—were more likely to audit a higher percentage of their transfusions.

Sixty percent (3924 of 5830) of the RBC transfusion episodes in the participants' review were for 1 unit, and 32% (2063 of 5830) were for 2 units at the same time. Single-unit RBC transfusions have become a centerpiece of blood management programs in recent years.¹²  Our data affirm both the feasibility of that approach and also the opportunity to expand it further. Conversely, the former taboo against 1-unit transfusions has largely been replaced, with only 9% (12 of 129) of facilities conducting such reviews in our study.

For participants at the time of this study (2008), the median review thresholds for most clinical indications were in the range of 8.0 to 8.9 g/dL Hgb. Those levels are somewhat higher than recent recommendations of 7.0 to 8.0 g/dL in stable patients, but some of the corroboration for those recommendations has appeared since our study was conducted.¹²  Patients with cardiovascular or pulmonary disease were ceded a 1-g rise in Hgb from the mentioned levels by the participants; for the subset of patients with acute cardiac events, the thresholds in use were reasonably aligned with recommendations for slightly higher Hgb levels.¹³ 

The involvement of physicians with transfusion expertise was a positive factor for audit criteria and for compliance with those criteria. Facilities with transfusion review thresholds of less than 10 g/dL Hgb were more likely to have a pathologist or transfusion medicine specialist actively involved in transfusion practice consultations (64% [53 of 83] versus 45% [19 of 42]). Better compliance with local criteria during the study period was associated with having hematology/oncology representation on the transfusion committee (79% versus 57% for those without).

There was an apparent discrepancy in practice compliance between the data submitted for the previous year and the study period of 50 recent transfusions examined. Across the participants, the institutional median percentage of RBC transfusions complying with facility guidelines during the previous year was 99.1% versus 69.0% [range, 0–100] in the study period. Perhaps the focused review process for the study was more stringent or more comprehensive than the routine review practice.

Intraoperative RBC transfusions presented a challenge for auditing. During the study period, slightly more than one-third (37%; 187 of 503) of the transfusions met institutional guidelines. However, only 31% (39 of 126) of the participants examined postoperative Hgb/Hct values for intraoperative transfusions. Using the rule of thumb of 1 g/dL increment of Hgb/unit transfused would permit extrapolation for what the Hgb would have been without transfusion.

During the Q-Probes reviews, 11.8% (764 of 5830) of RBC units were issued between midnight and 5:59 am. That corroborates data from the United Kingdom in which 13% to 20% of transfusions were given between 12:01 am and 8:00 am.¹⁴  In the Serious Hazards of Transfusion (SHOT) surveillance data from the United Kingdom, those morning hours were disproportionally associated with errors in picking up blood units from the transfusion service and administering them to patients.¹⁴ 

In conclusion, in this study of RBC transfusion audit practices in 128 hospitals, the institutional median values were 8.1 g/dL for pretransfusion Hgb and 69% for the proportion of transfusions meeting hospital criteria. Involvement of pathologists and transfusion medicine specialists in the review process correlated with stricter criteria and better compliance with those criteria. Red blood cell transfusion recommendations continue to evolve since this study was conducted. Comprehensive blood management programs emphasizing avoidance and nontransfusion correction of anemia as well as appropriate transfusion practice are becoming more common. Encouragingly, more than half of the RBC transfusion episodes in this study were for single units, providing frequent opportunities to reassess patients before giving a second or third unit.

We thank Christine Bashleben, MT(ASCP), senior technical specialist, Surveys, College of American Pathologists, Northfield, Illinois, for her invaluable assistance to the Q-Probe program.