The CRIT Study: Anemia and Blood Transfusion in the Critically Ill - Current Clinical Practice in the United States

Howard L. Corwin, MD; Andrew Gettinger, MD; Ronald G. Pearl, MD, PhD; Mitchell P. Fink, MD; Mitchell M. Levy, MD; Edward Abraham, MD; Neil R. MacIntyre, MD; M. Michael Shabot, MD; Mei-Sheng Duh, MPH, ScD; Marc J. Shapiro, MD

Disclosures

Crit Care Med. 2004;32(1) 

In This Article

Results

A total of 284 ICUs in 213 hospitals participated in the study. Hospital and ICU size are displayed in Table 1 . Of the 213 hospitals, 70% were characterized as urban, 26% suburban, and 4% rural. Of the ICUs, 31% were medical, 29% were surgical, and 40% were medical-surgical. Seventy-one percent of the ICUs were managed as "open" units. There was a full-time ICU director in 84% of the ICUs. Residents and fellows were present in 76% and 39% of ICUs, respectively. Only 19% of hospitals had an institutional transfusion protocol in place at the time of the study.

During the 9-month study period, a total of 4,892 patients were enrolled in the study. Patient characteristics are summarized in Table 2 . The mean age of the patients was 60 ± 18 yrs; 35% were >70 yrs old, and 30% were <50 yrs of age. The mean ICU LOS was 7.4 ± 7.3 days; 39% of patients remained in the ICU for ≤3 days, 36% stayed in the ICU for >1 wk, and 15% stayed for >2 wks. Admitting diagnostic categories and co-morbidities are shown in Table 3 , Table 4 , respectively. Sixty-one percent of patients required mechanical ventilatory support during their ICU stay, and 46% were mechanically ventilated at the time of ICU admission. The mean duration of mechanical ventilation was 7.1 ± 7.4 days for ventilated patients. ICU and hospital mortality rates for patients were 13% and 18%, respectively.

RBC transfusions performed in the ICU, after ICU discharge, and during the combined ICU and post-ICU periods are summarized in Table 5 . In total, 11,391 RBC units were transfused during the study period, including 9,990 RBC units in the ICU and 1,401 RBC units after ICU discharge. Overall, 44% of patients admitted to the ICU received one or more RBC units while in the ICU. The mean time to first transfusion was 2.3 ± 3.7 days (median, 1.0; lower quartile, 0.0; upper quartile, 3.0). More RBC transfusions were given in week 1; however, in subsequent weeks, patients received one to two RBC units per week (Fig. 1). Longer ICU stays were associated with both a higher percentage of patients transfused and more RBC units transfused per patient (Figure 2, Figure 3, respectively). Thirteen percent of patients received one or more RBC units after ICU discharge. Among these patients, 60% also received an RBC transfusion while in the ICU. Transfusion indications are shown in Table 12 . The most common indication (90%) reported was for low hemoglobin level.

Number of red blood cell (RBC) units transfused between day 1 and day 30. ANOVA, analysis of variance.

Percentage of patients receiving transfusions by intensive care unit (ICU) length of stay (LOS).

Mean number of red blood cell (RBC) units transfused by length of stay (LOS). ICU, intensive care unit.

Patients who were transfused had more total complications and were more likely to experience a complication ( Table 7 ). Approximately 4% of RBC transfusions were associated with a transfusion-related complication. The most common transfusion-related complications reported were fever (1.9%), fluid overload (1.7%), and hypotension (1%). The number of RBC units transfused was directly related to mortality. Mortality was 10% for patients who received no transfusions and 25% among patients who received six or more RBC units.

The pretransfusion hemoglobin levels are shown in Figure 4. The mean pretransfusion hemoglobin was 8.6 ± 1.7 g/dL. The same value pertained, irrespective of whether transfusions were given during the ICU stay or after discharge from the ICU ( Table 5 ). The pretransfusion hemoglobin was remarkably consistent across the range of ICU characteristics ( Table 8 ). The pretransfusion hemoglobin was also comparable for all transfusions (e.g., first, second, third) and ICU LOS. There were no clinically meaningful differences associated with age, sex, diagnostic category, APACHE II, or baseline SOFA score.

Pretransfusion hemoglobin.

The mean hemoglobin level at baseline was 11.0 ± 2.4 g/dL, and almost two thirds of patients had a baseline hemoglobin level of <12 g/dL. Hemoglobin level decreased throughout the duration of the study (Fig. 5). Individuals with a lower baseline hemoglobin level were more likely to receive an RBC transfusion. Almost 90% of patients with a baseline hemoglobin of ≤8 g/dL received an RBC transfusion. In contrast, only 20% of patients with a baseline hemoglobin of >12 g/dL received an RBC transfusion. Both time to first ICU transfusion (1.8 ± 1.7 days for baseline hemoglobin of ≤8 g/dL vs. 6.3 ± 6.2 days for baseline hemoglobin of >12 g/dL, p < .05) and total RBC units transfused (6.3 ± 7.1 units for baseline hemoglobin of ≤8 g/dL vs. 4.6 ± 4.4 units for baseline hemoglobin of >12 g/dL, p < .05) were significantly different between the low and high baseline hemoglobin groups.

Hemoglobin (Hb) levels day 1 through day 30. Trans/No Trans, ratio of patients who received transfusions to those who did not; ANOVA, analysis of variance.

Patients with a low baseline hemoglobin level (≤8 g/dL) presented with more hemodynamic instability, sepsis, and gastrointestinal bleeding, whereas patients with a higher baseline level (>12 g/dL) presented with more respiratory and cardiovascular problems ( Table 9 ). Similarly, a lower baseline hemoglobin level was associated with preexisting anemia or conditions often associated with anemia (e.g., cancer, renal disease), whereas a higher baseline hemoglobin level was associated with more pulmonary disease ( Table 10 ). Baseline hemoglobin levels of ≤10 g/dL were associated with higher APACHE II and SOFA scores and higher mortality and complication rates.

There was little association between age and either RBC transfusion rate (42% for patients of ≤50 yrs vs. 47% for patients of ≥70 yrs) or pretransfusion hemoglobin concentration (8.5 g/dL for patients of ≤50 yrs vs. 8.7 g/dL for patients of ≥70 yrs). In general, the incidence of RBC transfusion was relatively consistent across the co-morbidities; however, patients with preexisting anemia and hematologic disease tended to receive transfusions more frequently (58.3% and 58.7%, respectively), whereas patients with preexisting pulmonary disease tended to receive transfusions less frequently (37.5%). This pulmonary group tended to have a relatively higher proportion of patients with a baseline hemoglobin level of >12 g/dL and a lower proportion of patients with a baseline hemoglobin level of ≤8 g/dL ( Table 10 ). On the other hand, transfusions were much more frequent in patients with an admitting diagnosis of gastrointestinal hemorrhage (80%) and less frequent in patients admitted with a pulmonary (35%) or neurologic diagnosis (30%).

The mean baseline APACHE II and SOFA scores were 19.7 ± 8.2 and 6.2 ± 3.7, respectively. Both baseline APACHE II and SOFA score were significantly higher for patients with a baseline hemoglobin level of ≤10 g/dL. Low baseline APACHE II (≤15) and low baseline SOFA (≤6) scores were associated with a significantly decreased likelihood of RBC transfusion and fewer total RBC units transfused. There was a significant association between mean SOFA score during the course of the ICU stay and the number of RBC units transfused (no transfusions, SOFA scores 4-5; one to six RBC units, SOFA scores 6-7; more than six RBC units, SOFA scores 8-9).

The distribution of the age of RBC units transfused is shown in Figure 6. The mean age of all RBC units transfused was 21.2 ± 11.4 days. There was no difference in the age of RBC units whether transfused within or outside of the ICU, nor was there any difference in the age of RBC units transfused among different types of institutions or ICUs. There was no difference between the median age of the RBC units a patient received and any clinical outcome.

Age of red blood cell units transfused.

The number of RBC units transfused was significantly associated with increased ICU and hospital LOS compared with patients who did not receive transfusions ( Table 11 ). Patients with a transfusion amount of 1-2, 3-4, and >4 units had a corresponding increase in median ICU LOS of 2.1, 3.8, and 10.1 days, respectively; and an increase in median hospital LOS of 3.5, 6.7, and 16.6 days, respectively, as compared with the median ICU LOS of 4.6 days and hospital LOS of 11.0 days observed in the patients who did not receive transfusions. Baseline hemoglobin level was not statistically significantly associated with ICU or hospital LOS; however, a separate analysis shows that lower nadir hemoglobin levels were correlated with longer LOS.

RBC transfusion was also independently associated with higher mortality rates ( Table 12 ). Neither a baseline hemoglobin level of <10 g/dL nor a mean age of RBC units transfused of >2 wks was independently associated with an increase in mortality. However, in a separate model, nadir hemoglobin of <9.0 g/dL was associated with an increase in mortality. For the analysis using matching by propensity score to study mortality rate, 1,059 transfused patients (44.8%) were matched to 1,059 patients (41.8%) who did not receive transfusions. After adjusting for the propensity for receiving a blood transfusion, RBC transfusion remained statistically significantly associated with an increased risk for death (adjusted mortality ratio, 1.65; 95% confidence interval, 1.35-2.03; log-rank, p < .001) (Fig. 7).

Mortality by transfusion status for propensity-matched patients.

processing....