T
RANSFUSION
M
EDICINE
R
EVIEWS
JOURNAL CLUB
Richard Haspel, Jeannie Callum, and Sunny Dzik, Abstract Editors
Age of transfused blood: An independent predictor of
mortality despite universal leukoreduction. J.A. Weinberg,
G. McGwin Jr, R.L. Griffen et al. J Trauma 65:279-284,
2008
Both the effect of the age of blood and leukoreduction on
outcomes have been controversial. Weinberg et al combine these
2 hot topics in their analysis of trauma patients. The authors
report a retrospective analysis of trauma patients at their hospital
from 2000 through 2007. Of note, patients who died within 24
hours of admission were excluded. The number of units (all
leukoreduced) received during the first 24 hours was divided into
those that were less than 14 days old and those that were more
than 14 days old. The authors chose this number as they cite
several retrospective articles from a single group suggesting that
2 weeks may be an appropriate cutoff for the deleterious effects
of blood transfusion. The authors present odds ratios (ORs) for
the risk of death adjusted for age, sex, injury severity score (ISS),
mechanism of injury, total number of units in the first 24 hours,
and length of hospital stay. It appears that any adjustments made
for the number of units transfused were based on stratification
into categories and not based on dose as a continuous variable.
Although the number of units received in the first 24 hours
ranged from 1 to 74 (mean = 4.95), for unclear reasons, they
stratified the dose of blood into 3 groups: N1 unit, 2-3 units, and
N3 units.
Table 1 of the article is an attempt to describe baseline
variables of the different patients. Based on their stratification
scheme, the authors create 8 different groups of patients based on
the number of “old” or “young” blood they received. In the end,
they created a very confusing table with several important
baseline variables (ISS, units transfused in 24 hours) but also
missing many others (time period, other products infused,
products infused after 24 hours, mean age of transfused blood).
From this essentially univariate analysis, they noted differences
among the comparison groups regarding ISS, units transfused,
and mortality with patients receiving the most transfusions (N6)
not surprisingly doing the worst.
Table 2 is also based on the unexplained stratification into
categories and further contributed to my confusion as a reader.
As expected, patients receiving the most blood (N6 units),
whether young or old, had the highest ORs for death. The
authors, however, claim that because the size of the ORs was
greater in patients receiving less than 3 units of old vs young
blood (7.78 vs 3.79) that “blood stored beyond 2 weeks appears
to potentiate” the association of large volumes of blood with
mortality “despite a practice of universal leukoreduction.” It is
crucial to note that it appears the reference values used in
determining the ORs were different for the young blood and
old blood groups. For the effect of transfusion of young blood
and old blood, the reference ranges were patients who did not
receive any young blood or did not receive any old blood,
respectively. As such, with different reference values, it is not
appropriate to compare the ORs directly.
In a final subgroup of a subgroup analysis using the random
stratification pattern, patients who only received old blood and
received more than 3 units had a higher OR of death compared to
those who received only young blood. This value reached
borderline statistical significance (OR, 2.18; confidence interval,
1-4.97).
This article has numerous problems. Many baseline
variables are missing, and there was an apparently arbitrary
selection of stratification patterns in regard to number of units
received. In the end, given the subgroup analysis as well as
questionable statistical technique, few conclusions can be made
regarding any interaction between leukoreduction and the
storage lesion on outcomes. (RH)
Blood transfusion, anesthesia, surgery and risk of non-
Hodgkin lymphoma in a population-based case-control
study.. J.R. Cerhan, E.A. Engels, W. Cohen et al. Int J
Cancer 123:888-894, 2008
Several studies have linked transfusions, acting presumably
through an immunomodulatory effect, to increased risk of cancer.
In regard to non-Hodgkin lymphoma (NHL), the data are mixed.
The article by Cerhan et al is the latest additions to this body
of literature.
The authors used the Surveillance, Epidemiology, and End
Results cancer registry. Of 2248 newly diagnosed cased of NHL
from 1998 to 2000, the authors were eventually able to analyze
data from 759 cases. Controls were selected through random-
digit dialing or from Medicare eligibility files. Of the 2046
controls that the authors tried to contact, 529 were included in the
study. Participants were asked demographic questions and
occupational and pesticide use history. For transfusion history,
the question was “before 1 year, did you every have a blood
transfusion?” For any affirmative answer, further detail such as
the indication for transfusion, number and type of units, and age
at time of transfusion. Another major focus of the study was the
effect of surgery and anesthesia on NHL, so numerous questions
related to types of surgery were also asked. Transfusions were
categorized as ever/never, time since first transfusion (based on
prior studies: b5 years, 5-29 years, and 30+ years), and number
of transfusions and indication (grouped as trauma, obstetric,
surgical, or medical). For statistical analysis, adjustments were
made for study center, age, sex, and race. Although the results are
not presented, adjustments for educational level, body mass
index, and family history of lymphoma had no effect on the odds
ratios (ORs).
0887-7963/08/$ - see front matter
doi:10.1016/j.tmrv.2008.11.001
Transfusion Medicine Reviews,Vol23, No 1 (January), 2009: pp 75-82 75