By Michael Crawford, MD, Editor
SYNOPSIS: An observational study confirms the hypothesis that ischemic heart disease patients may do better with higher hemoglobin levels as compared to ICU patients without heart disease.
SOURCES: Ding YY, et al. Hemoglobin level and hospital mortality among ICU patients with cardiac disease who received transfusions. J Am Coll Cardiol 2015;66:2510-2518.
Rao SV, Vora AN. Transfusion in ischemic heart disease: Correlation, confounding, and confusion. J Am Coll Cardiol 2015;66:2519-2521.
The threshold hemoglobin (Hgb) level for red blood cell transfusion in hospitalized patients with cardiac disease is controversial. Thus, investigators from Boston University studied the Veterans Affairs (VA) electronic database to determine the Hgb level at which blood transfusion was associated with lower hospital mortality in medical ICU patients with cardiac disease. ICU admissions who had at least one transfusion in the first 30 days were the transfusion group; all others were the no transfusion group. The Hgb nadir was the lowest level before transfusion, or if not transfused the lowest level in the first 30 days in the ICU. Other variables included ICU admission diagnoses, comorbid conditions, and demographic characteristics. Researchers used adjusted linear regression analyses to analyze more than 5 years of ICU admissions data. Among the 258,826 ICU admissions, hospital death occurred in 12% and transfusions were noted in 12% during the first 30 days in the ICU. In addition to being older and sicker, those who died were twice as likely to have received a transfusion. In patients without cardiac disease, transfusion was associated with reduced adjusted hospital mortality when Hgb was < 7.7 g/dL. Above this level, transfusion was associated with higher mortality. In patients with cardiac disease, the corresponding level was 8.7 g/dL and 10 g/dL when the ICU admission diagnosis was acute myocardial infarction (MI). Sensitivity analyses in a smaller subset with more complete data showed that the Hgb levels below which mortality was reduced by transfusion could be about 1 g/dL lower than that for the total population. The authors concluded that in patients admitted to the ICU with comorbid cardiac disease, the Hgb level below which transfusion was associated with lower hospital mortality was < 8-9 g/dL and < 9-10 g/dL if the admitting diagnosis was acute MI.
COMMENTARY
Since the 1999 publication of the Transfusion Requirements in Critical Care (TRICC) randomized trial, critical care physicians and hospitals have pushed to restrict blood transfusions to those with a Hgb < 7 g/dL because this group showed lower mortality vs the comparison < 10 g/dL group. Cardiologists were concerned when this advice was applied to patients with acute ischemic heart disease (IHD) because the myocardium extracts nearly all the oxygen to it from the blood. The only way to deliver more oxygen to the myocardium is to deliver more oxygen, and a low Hgb would limit this. Some small observational studies supported this belief that higher Hgb thresholds for transfusion in IHD patients lowered mortality, but not all. Randomized, controlled trials (RCT) in acute coronary syndrome patients were called for. Two small pilot RCTs with a total of 155 patients showed conflicting results. It now seems unlikely that a large RCT will ever be conducted on this topic.
Given this background, this mega-observational analysis of more than 250,000 ICU admissions in the VA health system is of interest. It confirmed the TRICC study by showing that in ICU patients with no cardiac disease, the beneficial threshold was 7-8 g/dL, but in patients with cardiac comorbidities it was 8-9 g/dL and with acute MI it was 9-10 g/dL. Interestingly, both the American Association of Blood Banks and the American College of Cardiology/American Heart Association guidelines recommend a threshold of < 8 g/dL for patients with IHD. Thus, the weight of evidence and opinion seems to support higher Hgb thresholds for transfusion in IHD patients. The authors wisely suggested that there is probably a continuum of risk in ICU patients different from those with isolated medical disease, those with cardiac comorbidities, and those admitted with acute coronary syndrome. Also, some acute coronary syndrome patients may have severe medical illnesses, such as septic shock and pneumonia. Thus, considerable clinical judgment is required, and Hgb threshold levels are just guidelines.
There are significant limitations to this study. The accompanying editorial points out that the statistical techniques used are somewhat novel for an observational study and not fully vetted. Of course, the larger the study, the less likely all the details one would desire are present. For example, we do not know the do-not-resuscitate status of the patients. Also, only 3% of the study population is women, but that represents about 7000 individuals. In addition, the data analyzed is from 2001-2005. Newer concepts and therapies may have altered ICU care in the last 10 years. Finally, this study does not shed any light on heart failure patients, which is an even more complex situation. For now, the transfusion Hgb threshold for IHD patients should increase at least 1 g/dL to < 8 and perhaps higher, especially for acute MI patients.
