Secondary Logo

Journal Logo

ORIGINAL ARTICLES

Age of red blood cells is not associated with in-hospital mortality in massively transfused patients

Saadah, Nicholas H. PhD; Wood, Erica M. MD; Bailey, Michael J. PhD; Cooper, D. James MD; French, Craig J. MD; Haysom, Helen E. BS; Sparrow, Rosemary L. PhD; Wellard, Cameron J. PhD; McQuilten, Zoe K. MD, PhD; On behalf of the Australian and New Zealand Massive Transfusion Registry Steering Committee

Author Information
Journal of Trauma and Acute Care Surgery: August 2021 - Volume 91 - Issue 2 - p 279-286
doi: 10.1097/TA.0000000000003192

Abstract

Red blood cells (RBCs) collected for transfusion may be refrigerated for several weeks, typically up to 42 days, before infusion.1 During storage, RBCs undergo biochemical and biophysical changes, referred to as the storage lesion.2–4 Investigating the clinical consequences of the storage lesion has been an active area of research, particularly over the past two decades; at least 14 large (>1,000 adult patients) observational studies5–18 and 4 large multicenter “age of blood” clinical trials19–22 in adult patients that explored a potential association between storage time of transfused RBCs and adverse patient outcomes have been published.

While observational studies have reported mixed results, the four large trials all showed no correlation between storage time and mortality or change in Multiple Organ Dysfunction Score.23 Two of the trials (age of blood evaluation, TRANSfusion versus Fresher red-cell USE) were in critically ill intensive care patients; one (REd CEll Storage duration Study) in complex cardiac surgery patients, with change in Multiple Organ Dysfunction Score as the primary outcome; and one trial (INforming Fresh versus Old Red cell Management) included any patient requiring RBC transfusion. All four trials included patients requiring one or more RBC units, meaning the median number of RBC units transfused was relatively small, between 2 and 4 U.19–22

The question therefore remains whether similar results are observed in patients transfused with larger numbers of RBCs, such as critically bleeding patients requiring massive transfusion (MT). Life-threatening critical bleeding can occur in a wide range of clinical contexts, from gastrointestinal and obstetric hemorrhage, to surgery and trauma. The combination of the inherent complexities of clinical management and the logistical issues of providing blood products for critically bleeding patients make it difficult to conduct a randomized age of blood clinical trial in this patient cohort. To our knowledge, no observational studies have specifically explored an association between storage time and mortality in critically bleeding/massively transfused patients, although at least three observational studies on trauma patients have reported a positive correlation between RBC storage time and mortality.14,24,25 We hypothesized that any clinical effects of the RBC storage lesion may be accentuated in critically bleeding patients receiving MT.

PATIENTS AND METHODS

The Australian and New Zealand Massive Transfusion Registry (ANZ-MTR) is a clinical data set of adult patients (≥18 years) who had critical bleeding requiring MT (≥5 RBC units in any 4-hour period during hospital admission, any bleeding context).26 We used data from this registry to test the association between in-hospital mortality and storage time in Australian and New Zealand adult MT patients.

Patients

All MT cases were identified using the hospital laboratory information system at the 29 hospitals participating in the ANZ-MTR at the time of data extraction. Massive transfusion was defined as ≥5 RBC units issued within any 4-hour period during the hospital admission. Patients with all causes of bleeding were included. Detailed information on the ANZ-MTR methods has been previously published.26

Red Cell Units

Both Australia and New Zealand have national blood services and primarily use the whole blood buffy-coat depletion method to produce their RBC inventories.27,28 During the study period, 2007 to 2018, all RBC units were prestorage-leukodepleted and stored in saline-adenine-glucose-mannitol additive solution. Because maximum storage time for RBCs is 35 days in New Zealand and 42 days in Australia, the data for each country were analyzed separately to avoid the bias introduced by this difference.

Data Sources

For each MT case, the date/time of issue for all transfused blood products issued by the hospital transfusion laboratory was extracted, along with the unique registry case ID, patient age at admission, patient sex, admission International Classification of Diseases, Tenth Revision, Australian Modification) diagnostic and Australian Classification of Health Interventions procedure codes, hospital length of stay, length of time in intensive care unit, MT year, lactate measurement taken closest to MT start (as a proxy for shock), country, and in-hospital mortality. Because returned blood product units are registered as such, all units ultimately cataloged as “issued” were in fact transfused. Massive transfusion cases for which in-hospital mortality and storage time data for all transfused RBC units were available were included for analysis.

Storage time for each RBC unit was calculated using the unit expiry date (unit storage time = maximum storage time (days) − unit expiry date + unit issue date/time where maximum storage time = 43 days [Australia], 36 days [New Zealand]), when available, with the unit production date (unit storage time = unit production date + maximum storage time (days) − unit issue date/time) being used otherwise. Note that, because both national blood services set RBC unit expiry at 11:59:59 pm of the unit expiry date, which is itself the 42nd (Australia) or 35th (New Zealand) day after blood donation, 43 (Australia) and 36 days (New Zealand) were used as the maximum storage time. Charlson Comorbidity Index (CCI) was calculated for each patient based on their diagnostic codes.29 Bleeding context was categorized using the diagnostic and procedure codes into 10 categories: (1) trauma; (2) cardiothoracic surgery, (3) vascular surgery, (4) cardiothoracic surgery plus vascular surgery, (5) liver transplant, (6) gastrointestinal hemorrhage, (7) gastrointestinal hemorrhage plus surgical bleed, (8) other surgery, (9) medical/other, and (10) obstetric hemorrhage. Details on the categorization of bleeding contexts have been previously published.30

Statistical Methods

Our methodological design was based on the expert commentary provided by the Biomedical Excellence for Safer Transfusion Collaborative that addressed the methodological pitfalls associated with this research question.31 As such, three analyses were performed:

Analysis 1

Logistic regression with in-hospital mortality as the outcome and mean storage time of the transfused RBCs (STmean, as quartiles), number of RBCs, patient age, male sex, year of MT, and CCI as predictors, and treating center modeled as a random effect was run on each primary bleeding context separately. Number of transfused RBCs was treated as a continuous variable, with patients receiving 30 or more RBC units analyzed as a category of 30+ RBCs. The 30+ RBCs category was necessary, as the linear relationship assumed by logistic regression between number of RBCs and the logit of our outcome, mortality, breaks down for patients receiving extremely high numbers of RBCs. We checked this relationship, as well as the relationship between mortality and year, via visual inspection of scatter plots. Results are reported as odds ratios (ORs) comparing mortality for each of the mean storage time quartiles with the first (reference) quartile of RBCs with the shortest mean storage time (ORQ2vQ1, ORQ3vQ1, and ORQ4vQ1). A power calculation accompanies the primary analysis to estimate the minimum detectable effect size.

Analysis 2

Logistic regression with in-hospital mortality as the outcome and proportion of each patient’s transfused RBCs that were 30 days or older (propOLD, as a continuous predictor), number of RBCs, patient age, male sex, year of MT, and CCI as predictors, and treating center modeled as a random effect was run on each primary bleeding context separately. The result was reported as the OR describing the effect on mortality of an increase of one in the proportion of RBC units stored for more than 30 days.

Analysis 3

Logistic regression with in-hospital mortality as the outcome and the scalar age of blood index (SBI), as introduced by DeSantis and colleagues,25 number of RBCs, patient age, male sex, year, and CCI as predictors, and treating center modeled as a random effect was run on each primary bleeding context separately. DeSantis and colleagues’ methodology requires choosing thresholds separating the storage time range into four “quartiles”—we used 0 to 10, 10 to 20, 20 to 30, and 30+ days. The result was reported as the OR describing the effect on mortality of an increase of one in the SBI.

For the STmean primary predictor, quartiles were tabulated and compared with regard to in-hospital mortality, number of RBCs transfused, patient age, patient sex, CCS, mean RBC storage time, post-MT hospital length of stay (for patients surviving their admission), post-MT time to death (for patients dying in hospital), lactate measurement closest to MT start time (mmol/L), fresh frozen plasma units/RBC units ratio, and platelet units/RBC units ratio. A one-way analysis of variance was run to test the assumption that the four quartiles did not differ significantly with regard to these values, and the residuals for each were plotted to check the assumption of normal distribution. Continuous variables are reported as medians (interquartile range [IQR]), proportions are reported as frequency (percentages). A two-sided p value was calculated at the α = 0.05 level to test statistical significance throughout the analysis. All calculations were carried out in MATLAB (R2016b; The MathWorks, Inc., Narick, MA).

RESULTS

MT Cohort Characteristics and Transfused RBCs

Data were available for a total of 9,237 MT cases from 29 hospitals participating in the ANZ-MTR during all or part of the study period of 2007 to 2018. Figure 1 shows the MT cohort selection scheme. After eliminating cases for which storage time data for all transfused RBCs were not available (537 cases) and cases with unknown in-hospital mortality (15 cases), 8,685 cases (94% of cases in the ANZ-MTR), involving transfusion of a total of 126,622 RBC units, were analyzed (Fig. 1).

Figure 1
Figure 1:
Flow schema of MT study cohort selection.

Figure 2 shows histograms displaying the distributions of RBC unit storage times for both Australia and New Zealand. For the Australian cohort, the median number of RBC units transfused was 11 (IQR, 8–17), and the median age of transfused RBC units was 18.6 days (range, 1.5–43.0 days). For the New Zealand cohort, the median number of RBC units transfused was 10 (IQR, 7–15), and the median age of transfused RBC units was 15.8 days (range, 1.6–36.0 days).

Figure 2
Figure 2:
Histograms of storage time for RBC units transfused for (A) Australia and (B) New Zealand data during massive transfusion cases. (A) Australian data (100,453 U). (B) New Zealand data (26,169 U).

Table 1 shows descriptive characteristics for the Australian and New Zealand cohorts and for each of the four STmean quartile subcohorts. The mean storage times for the four STmean quartiles were as follows: (Australia) Q1 STmean = 12.5 days, Q2 STmean = 17.7 days, Q3 STmean = 22.3 days, and Q4 STmean = 29.8 days; (New Zealand) Q1 STmean = 11.3 days, Q2 STmean = 15.3 days, Q3 STmean = 18.7 days, and Q4 STmean = 24.5 days. Supplementary Figure 1 (http://links.lww.com/TA/B979) displays the residuals for the variables in Table 1. Given that most were not normally distributed, we ran a Kruskal-Wallis one-sided analysis of variance. The quartiles are similar, albeit with the differences between quartiles for some variables, such as fresh frozen plasma/RBC and platelet/RBC unit ratios, reaching statistical significance. In these cases, the differences are small and not likely to be clinically significant.

TABLE 1 - Descriptive Characteristics for the MT Cases for the Entire Cohort and Each of the Four Mean Storage Time (STmean) Quartiles
Entire Cohort STmean Q1 STmean Q2 STmean Q3 STmean Q4 Sig.
Australian cohort
 Transfusion episodes 6,716 1,679 1,679 1,679 1,679
 Mean STmean (range), d 20.6 (3.1–41.7) 12.5 (3.1–15.5) 17.7 (15.5–19.9) 22.3 (19.9–24.9) 29.8 (24.9–41.7) p < 0.01
 In-hospital mortality 0.21 0.21 0.23 0.20 0.18 p < 0.01
 RBCs, mean (SD) 15.0 (13.1) 15.5 (14.0) 16.0 (14.4) 14.7 (11.7) 13.6 (11.8) p < 0.01
 Patient age, y 61.0 (46.0–73.0) 61.0 (45.3–73.0) 61.0 (45.0–73.0) 62.0 (46.0–73.0) 62.0 (46.0–74.0) p = 0.32
 Proportion male 0.64 0.64 0.64 0.65 0.64 p = 0.84
 CCI 2 (0–4) 2 (0–4) 1 (0–4) 1 (0–4) 2 (0–4) p = 0.99
 Hospital post-MT LOS, d 16.6 (8.8–33.0) 15.0 (8.1–28.1) 17.5 (8.7–34.0) 18.0 (9.3–37.8) 16.5 (8.8–32.0) p = 0.06
 Post-MT time to death 2.2 (0.5–10.3) 2.2 (0.6–9.8) 2.6 (0.6–12.3) 2.4 (0.4–11.8) 1.5 (0.4–8.5) p = 0.79
 Lactate, mmol/L 2.3 (1.3–4.4) 2.2 (1.3–4.3) 2.4 (1.4–4.8) 2.4 (1.4–4.4) 2.3 (1.3–4.3) p = 0.09
 FFP/RBC ratio 0.5 (0.3–0.8) 0.5 (0.3–0.8) 0.5 (0.3–0.8) 0.5 (0.3–0.8) 0.4 (0.2–0.7) p < 0.01
 plts/RBC ratio 0.1 (0.0–0.2) 0.1 (0.0–0.2) 0.1 (0.0–0.2) 0.1 (0.0–0.2) 0.1 (0.0–0.2) p < 0.01
New Zealand cohort
 Transfusion episodes 1,969 492 492 493 492
 Mean STmean (range), d 17.4 (3.6–35.6) 11.3 (3.6–13.7) 15.3 (13.7–16.8) 18.7 (16.8–20.7) 24.5 (20.7–35.6) p < 0.01
 In-hospital mortality 0.25 0.30 0.23 0.24 0.22 p = 0.03
 RBCs, mean (SD) 13.3 (10.6) 15.1 (13.6) 13.7 (10.6) 13.1 (9.7) 11.4 (7.4) p < 0.01
 Patient age, y 63.0 (47.0–73.0) 62.0 (48.5–73.0) 63.0 (46.5–73.0) 63.0 (48.0–73.0) 62.0 (42.0–74.0) p = 0.81
 Proportion male 0.59 0.53 0.60 0.58 0.64 p = 0.01
 CCI 1 (0–3) 1 (0–3) 1 (0–3) 1 (0–3) 1 (0–3) p = 0.79
 Hospital post-MT LOS, d 10.5 (6.5–19.5) 11.5 (7.1–19.7) 11.0 (6.9–20.0) 10.9 (6.3–20.3) 9.1 (6.0–16.6) p = 0.03
 Post-MT time to death 1.8 (0.5–6.4) 2.0 (0.6–6.9) 2.3 (0.5–7.8) 1.3 (0.4–6.2) 1.8 (0.4–6.0) p = 0.34
 lactate, mmol/L 3.9 (1.8–3.9) 3.7 (1.6–3.9) 3.5 (1.7–3.9) 3.9 (1.9–3.9) 3.9 (1.9–3.9) p = 0.05
 FFP/RBC ratio 0.5 (0.3–0.7) 0.5 (0.3–0.8) 0.5 (0.3–0.7) 0.4 (0.3–0.7) 0.4 (0.3–0.7) p < 0.01
 plts/RBC ratio 0.1 (0.0–0.3) 0.3 (0.1–0.4) 0.1 (0.0–0.3) 0.1 (0.0–0.2) 0.1 (0.0–0.2) p < 0.01
Unless otherwise indicated, results are medians (IQRs). p Values presented are those for one-way analysis of variance analysis of each characteristic across the four STmean quartiles.
FFP, fresh frozen plasma; ICU, intensive care unit; LOS, length of stay; plts, platelets; Q#, quartile #; sig., significance; ST, storage time.

Based on a 20% mortality rate, with a minimum of 6,600 patients, this study will have a >90% power (two-sided p value of 0.05) to detect a difference in the mean age of blood between survivors and nonsurvivors equal to 10% of one SD. The SD in mean storage time was 10.2 days, suggesting that our study was powered to detect if a 1-day difference in storage time resulted in statistically significantly different mortality.

Logistic Regression Analyses

Given our methodological choice to enter only bleeding contexts with at least 100 patients, into our logistic regression analysis, we ran our analysis on trauma, cardiovascular surgery, vascular surgery, gastrointestinal hemorrhage, and surgical (other) cohorts for both Australia and New Zealand data, and, additionally, liver transplant and medical (other) for the Australian cohort. Table 2 shows the results of the logistic regression analyses. Note that, for all three analyses, the obstetric hemorrhage cohort was excluded from analysis as only one death occurred in this cohort, yielding too few outcomes for statistical analysis. Supplementary Figures S2 (http://links.lww.com/TA/B979) and S3 (http://links.lww.com/TA/B979) display the relationship between mortality and both (S2) number of RBCs and (S3) year, showing the linear relationships assumed by our logistic regression model.

TABLE 2 - Outcomes of Logistic Regression Analyses Modeling Number of RBC Units, Patient Age, Male Sex, Year of MT, and CCI as Predictors, and Treating Center Modeled as a Random Effect, Run on Each Bleeding Context for Which at Least 100 Episodes Were Observed
Episodes ORQ4vQ1 Sig. ORpropOLD Sig. ORsbi Sig.
(a) Australian Data
 Australian cohort
 Trauma 1,630 0.89 (0.64–1.24) p = 0.49 1.32 (0.81–2.13) p = 0.26 1.02 (0.92–1.14) p = 0.67
 Cardiothoracic surgery 1,283 0.83 (0.56–1.23) p = 0.37 0.77 (0.43–1.37) p = 0.38 0.96 (0.86–1.08) p = 0.50
 Vascular surgery 519 0.64 (0.37–1.11) p = 0.11 0.98 (0.90–1.07) p = 0.61 0.93 (0.79–1.09) p = 0.36
 Liver transplant 303 1.10 (0.21–5.81) p = 0.91 0.20 (0.00–8.87) p = 0.40 0.96 (0.57–1.63) p = 0.88
 Gastrointestinal hemorrhage 1,243 0.73 (0.49–1.08) p = 0.11 0.61 (0.35–1.06) p = 0.08 0.91 (0.81–1.02) p = 0.10
 Surgical (other) 1,270 0.68 (0.39–1.21) p = 0.20 0.83 (0.37–1.87) p = 0.65 0.90 (0.76–1.06) p = 0.22
 Medical/other 200 1.16 (0.49–2.75) p = 0.74 1.09 (0.40–2.96) p = 0.86 0.99 (0.79–1.22) p = 0.90
(b) New Zealand data
 New Zealand cohort
 Trauma 194 0.86 (0.34–2.16) p = 0.75 0.45 (0.06–3.49) p = 0.44 0.96 (0.64–1.46) p = 0.88
 Cardiothoracic surgery 598 0.80 (0.46–1.38) p = 0.42 0.25 (0.03–2.43) p = 0.23 0.89 (0.75–1.05) p = 0.17
 Vascular surgery 279 1.47 (0.68–3.14) p = 0.32 1.06 (0.24–4.59) p = 0.94 1.20 (0.89–1.62) p = 0.22
 Gastrointestinal hemorrhage 270 0.81 (0.32–2.08) p = 0.67 0.88 (0.14–5.39) p = 0.89 0.92 (0.61–1.41) p = 0.73
 Surgical (other) 381 0.59 (0.22–1.56) p = 0.29 0.34 (0.03–3.63) p = 0.37 0.69 (0.46–1.05) p = 0.08
Note that for analysis 1 (STmean as predictor), only STmean Q4 versus Q1 is here presented. Full results for this along with outcomes for all covariates are presented in Supplementary Table S1 (http://links.lww.com/TA/B979).
ORpropOLD, odds ratio describing change in mortality accompanying increase of 1 in proportion of RBCs older than 30 days; ORq4vq1, odds ratio comparing mortality for mean storage time fourth quartile versus first quartile; ORsbi, odds ratio describing change in mortality accompanying increase of 1 in DeSantis and colleagues' scalar blood index; Sig., significance.

Analysis 1 (STmean as Predictor)

Comparing each quartile to Q1 (reference quartile) by logistic regression for each bleeding context separately while correcting for number of RBCs, patient age, male sex, year, and CCI, none of the quartiles yielded a statistically significant OR (Table 2 shows Q4 vs. Q1 results; Supplementary Table 1 [http://links.lww.com/TA/B979] shows full logistic regression results).

Analysis 2 (propOLD as Predictor)

In our second analyses, this one using propOLD as a continuous predictor, propOLD was not a statistically significant predictor of in-hospital mortality in any bleeding context, in any of the Australian or Kiwi cohorts.

Analysis 3 (DeSantis et al.’s SBI as a Predictor)

In our third analyses, this one using our specified SBI as a continuous predictor and correcting, SBI was not a statistically significant predictor of in-hospital mortality in any the bleeding contexts in either the Australian or Kiwi cohorts.

DISCUSSION

In this observational study of 8,685 massively transfused adult patients, we found no systematic correlation between storage time of transfused RBC units and in-hospital mortality. Our methodology was designed to avoid the pitfalls common to observational studies addressing this research question31 by using three different storage time measures (i.e., mean storage time of each patient’s transfused RBCs, proportion of each patient’s transfused RBCs older than 30 days, and each case’s SBI) and analyzing each bleeding context separately. Our results are consistent with the findings from the four recent large-scale randomized clinical trials that addressed the potential association between age of blood and mortality.19–22 However, all of these trials used only dichotomized storage time variables (i.e., fresher vs. longer-stored RBCs), and none had a significant cohort of massively transfused patients.

Supplementary Figure S4 (http://links.lww.com/TA/B979) in the online supplement is a video (http://links.lww.com/TA/B985) file showing a three-dimensional plot of number of RBCs versus average storage time versus mortality. For this graphic, the number of RBCs and average storage time are each apportioned into six categories (sextiles). The video format allows this 3-dimensional surface plot to be visualized from various angles, clearly showing both the clear association between number of transfused RBCs (covariate) and mortality (outcome), and the lack of a clear association between average storage time (independent variable) and mortality. Similar surface plots for proportion of RBCs ≥30 days old and for SBI as the independent variables likewise show no association (data not shown).

As pointed out by Trivella and colleagues,32 using dichotomized groups for comparison is appropriate when a linear relationship between the predictor and the outcome variable can be expected. However, the assumption of a linear relationship between storage time and mortality is one without an evidence base and should therefore not be assumed.32 Our use of quartiles to define storage time groups allows evaluation of this potential association across the range of storage times within the MT patient cohort.

Furthermore, our use of three different metrics to define the storage time lesion, as suggested by van de Watering and colleagues,31 controls for the realities of correlation testing in patients receiving multiple RBC units, often of very different storage times. A potential association between this storage lesion and storage time could take on any number of multiple mathematical forms (e.g., linear relationship, nonlinear relationship, binary relationship), complicating analysis designed to estimate the association. The three measures we chose—mean storage time of transfused RBCs, proportion of transfused RBCs ≥30 days, and DeSantis and colleagues’ SBI—control for three different mathematical forms. Figure 3 illustrates this by showing these storage time measures for four theoretical patients. Supposing that the RBC storage lesion is linearly associated with storage time, a patient’s total storage lesion exposure would be measured by the sum total of the transfused RBC units’ storage times. Our analysis using mean storage time of the transfused RBC units, which is this total scaled by the number of RBCs, while simultaneously modeling the number of RBCs as a covariate is thus a test of a potential linear association between storage lesion and storage time. However, Figure 3 shows how patients transfused with RBCs of very different storage times can have the same mean storage time, which poses a problem should the relationship between RBC storage lesion and storage time be nonlinear (e.g., exponential, periodic).

Figure 3
Figure 3:
Hypothetical schema of four patients’ transfusion cases. Depending on the storage time measure used (e.g., mean storage time, maximum storage time, proportion of RBC units older than a given threshold), transfusion cases with RBCs of vastly different storage times can have identical storage time measures. Numbers shown on each RBC unit’s icon indicate storage time (days). prop. ≥30d, proportion of RBC units stored for ≥30 days before transfusion; ST, storage time (days) of RBC units.

DeSantis and colleagues’ SBI takes the distribution of the RBC units’ storage times into account when calculating the storage lesion, thus accounting for the possibility of a nonlinear association between storage lesion and storage time. Note in Figure 3 how patients 1, 2, and 4 have different SBIs despite having identical mean storage times. However, like mean storage time, SBI assumes a steadily, if not linearly, increasing storage lesion with storage time. Including an analysis using proportion of RBCs ≥30 days tests for a binary association between storage lesion and storage times above a critical threshold. Returning to Figure 3, note how patients 2 and 3 have identical mean storage times and SBIs but different proportions of RBCs older than 30 days. Given our incomplete understanding of the biomechanisms dictating the progression of the RBC storage lesion and its relation to storage time, we believe that the use of storage time measures based on different mathematical forms strengthens our analysis.

Strengths and Limitations

To our knowledge, our study represents the first large-scale observational study of adult MT patients not limited to the trauma setting. The strengths of our study include its large sample size, the large volume of RBCs received by observed patients, and the fact that cases were collected from many (29) sites. In addition, given that in-hospital mortality is our outcome, the high (21.7%) mortality of the analyzed patients increased the statistical power of our study. Finally, because these 29 sites are located in only two countries whose data we analyzed separately, our analysis benefits from a high level of consistency in blood collection, processing, storage, and transfusion among the sites analyzed together. This is in contrast to multicenter studies that sourced blood components from different blood product providers and different countries.

Our study was limited by the data sent to the ANZ-MTR that provided time of issue of each RBC unit, rather than exact time of transfusion. In the circumstance of MT, we felt it reasonable to assume a unit of blood was transfused shortly following its issue; however, this may not always be the case. We did not take into account when during the hospital admission the MT took place. While different times within a hospital admission represent different situations with regard to patient criticality, our analyzing of each bleeding context separately partially addresses this limitation. Given our lack of data on potential confounders specific to bleeding contexts, such as Injury Severity Score for trauma patients, we were not able to correct for these. In addition, given that transfusion analyses are subject to truncation by death, a survival analysis would have here been a preferable methodological approach. However, because we have access only to time of issue, rather than time of transfusion, this was not here a possibility.

The rate of accumulation of lesions during storage of RBC units can vary between donors.3,33 We have no data on the blood donors, nor quality measures, of the transfused RBCs; therefore, the potential impact of donor-related factors on variability of the quality of stored RBCs could not be assessed. Given that our study cohort of massively transfused patients received a minimum of five RBC units, each from an individual donor, it is reasonable to assume that the various factors relevant to variability between donors were randomly distributed and thus would not bias the results.

Despite the efforts taken to ensure robust methodology, our statistical methods remain subject to potential bias from two sources. First, our analysis did not consider the order in which RBCs were transfused, meaning two patients receiving RBCs of the same age but in different orders were treated as equivalent. Second, the dichotomous nature of two of our three storage time measures, propOLD and SBI, mean that patients transfused with RBCs of similar ages may end up with divergent propOLD and SBI values, as demonstrated by comparing patients 2, 3, and 4 in Figure 3.

Conclusion

In a large cohort of massively transfused adult patients, we found no systematic association between storage time of RBCs and in-hospital mortality. Furthermore, we found no evidence that transfusion with RBCs 30 days or older was associated with increased risk of death. These findings are in keeping with recently published randomized clinical trials in critical care and general hospital patients and support the current usual transfusion inventory management practice of issuing oldest RBCs for transfusion first.

AUTHORSHIP

N.H.S. designed and performed the analysis, interpreted the data, and was the primary author of the article. E.M.W. contributed in the writing of and critical revision of the article. M.J.B. consulted on the statistical methodology and contributed to the analysis and interpretation of the data. D.J.C. contributed in the writing of and critical revision of the article. C.J.F. contributed in the writing of and critical revision of the article. H.E.H. was responsible for data acquisition and was involved in the writing of and critical revision of the article. R.L.S. was responsible for data acquisition and was involved in the writing of and critical revision of the article. C.J.W. consulted on the statistical methodology and contributed in the analysis and interpretation of the data. Z.K.M. contributed in the design of the study and the interpretation of the data, along with the writing of and critical revision of the article.

ACKNOWLEDGMENT

We thank the staff at the participating hospitals involved in providing data to the ANZ-MTR. We acknowledge, with thanks, the leadership of the ANZ-MTR Steering Committee and the support of the organizations that fund the registry.

DISCLOSURE

The authors declare no conflicts of interest. The ANZ-MTR has received funding from the National Health and Medical Research Council (Australia), the National Blood Authority (Australia), the Victorian Government Department of Health and Human Services, the Australian Red Cross Blood Service, the New Zealand Blood Service, CSL Behring, and Monash University. Z.K.M. is the recipient of a 2016 4-year National Health and Medical Research Council Early Career Fellowship (APP1111485).

REFERENCES

1. Sparrow RL. Time to revisit red blood cell additive solutions and storage conditions: a role for “omics” analyses. Blood Transfus. 2012;10(Suppl 2):s7–s11.
2. Putter JS, Seghatchian J. Cumulative erythrocyte damage in blood storage and relevance to massive transfusions: selective insights into serial morphological and biochemical findings. Blood Transfus. 2017;15(4):348–356.
3. D’Alessandro A, Zimring JC, Busch M. Chronological storage age and metabolic age of stored red blood cells: are they the same?Transfusion. 2019;59(5):1620–1623.
4. Yoshida T, Prudent M, D’alessandro A. Red blood cell storage lesion: causes and potential clinical consequences. Blood Transfus. 2019;17(1):27–52.
5. Van De Watering L, Lorinser J, Versteegh M, Westendord R, Brand A. Effects of storage time of red blood cell transfusions on the prognosis of coronary artery bypass graft patients. Transfusion. 2006;46(10):1712–1718.
6. Koch CG, Li L, Sessler DI, Figueroa P, Hoeltge GA, Mihaljevic T, Blackstone EH. Duration of red-cell storage and complications after cardiac surgery. N Engl J Med. 2008;358(12):1229–1239.
7. Aubron C, Bailey M, McQuilten Z, Pilcher D, Hegarty C, Martinelli A, Magrin G, Irving D, Cooper DJ, Bellomo R. Duration of red blood cells storage and outcome in critically ill patients. J Crit Care. 2014;29(3):476.e1–476.e8.
8. Min JJ, Bae JY, Kim TK, Hong DM, Hwang HY, Kim KB, Han KS, Jeon Y. Association between red blood cell storage duration and clinical outcome in patients undergoing off-pump coronary artery bypass surgery: a retrospective study. BMC Anesthesiol. 2014;14:95.
9. Heddle NM, Eikelboom J, Liu Y, Barty R, Cook RJ. Exploratory studies on the age of transfused blood and in-hospital mortality in patients with cardiovascular diagnoses. Transfusion. 2015;55(2):364–372.
10. Goel R, Johnson DJ, Scott AV, Tobian AA, Ness PM, Nagababu E, Frank SM. Red blood cells stored 35 days or more are associated with adverse outcomes in high-risk patients. Transfusion. 2016;56(7):1690–1698.
11. Weinberg JA, McGwin G Jr., Griffin RL, Huynh VQ, Cherry SA 3rd, Marques MB, Reiff DA, Kerby JD, Rue LW 3rd. Age of transfused blood: an independent predictor of mortality despite universal leukoreduction. J Trauma. 2008;65(2):279–282.
12. Eikelboom JW, Cook RJ, Liu Y, Heddle NM. Duration of red cell storage before transfusion and in-hospital mortality. Am Heart J. 2010;159(5):737–743.e1.
13. Edgren G, Kamper-Jørgensen M, Eloranta S, et al. Duration of red blood cell storage and survival of transfused patients (CME). Transfusion. 2010;50(6):1185–1195.
14. Weinberg JA, McGwin G Jr., Vandromme MJ, Marques MB, Melton SM, Reiff DA, Kerby JD, Rue LW III. Duration of red cell storage influences mortality after trauma. J Trauma. 2010;69(6):1427–1431.
15. van Straten AH, Soliman Hamad MA, van Zundert AA, Martens EJ, ter Woorst JF, de Wolf AM, Scharnhorst V. Effect of duration of red blood cell storage on early and late mortality after coronary artery bypass grafting. J Thorac Cardiovasc Surg. 2011;141(1):231–237.
16. McKenny M, Ryan T, Tate H, Graham B, Young VK, Dowd N. Age of transfused blood is not associated with increased postoperative adverse outcome after cardiac surgery. Br J Anaesth. 2011;106(5):643–649.
17. Middelburg RA, van de Watering LM, Briët E, van der Bom JG. Storage time of red blood cells and mortality of transfusion recipients. Transfus Med Rev. 2013;27(1):36–43.
18. Kadar A, Chechik O, Katz E, Blum I, Meghiddo G, Salai M, Steinberg E, Sternheim A. The effects of “old” red blood cell transfusion on mortality and morbidity in elderly patients with hip fractures—a retrospective study. Injury. 2013;44(6):747–750.
19. Lacroix J, Hébert PC, Fergusson DA, et al. Age of transfused blood in critically ill adults. N Engl J Med. 2015;372(15):1410–1418.
20. Steiner ME, Ness PM, Assmann SF, et al. Effects of red-cell storage duration on patients undergoing cardiac surgery. N Engl J Med. 2015;372(15):1419–1429.
21. Heddle NM, Cook RJ, Arnold DM, et al. Effect of short-term vs. long-term blood storage on mortality after transfusion. N Engl J Med. 2016;375(20):1937–1945.
22. Cooper DJ, McQuilten ZK, Nichol A, et al. Age of red cells for transfusion and outcomes in critically ill adults. N Engl J Med. 2017;377(19):1858–1867.
23. McQuilten ZK, French CJ, Nichol A, Higgins A, Cooper DJ. Effect of age of red cells for transfusion on patient outcomes: a systematic review and meta-analysis. Transfus Med Rev. 2018;32(2):77–88.
24. Jones AR, Patel RP, Marques MB, et al. Older blood is associated with increased mortality and adverse events in massively transfused trauma patients: secondary analysis of the PROPPR trial. Ann Emerg Med. 2019;73(6):650–661.
25. DeSantis SM, Brown DW, Jones AR, Yamal JM, Pittet JF, Patel RP, Wade CE, Holcomb JB, Wang H; PROPPR Study Group. Characterizing red blood cell age exposure in massive transfusion therapy: the scalar age of blood index (SBI). Transfusion. 2019;59:2699–2708.
26. Oldroyd JC, Venardos KM, Aoki NJ, et al. Improving outcomes for hospital patients with critical bleeding requiring massive transfusion: the Australian and New Zealand Massive Transfusion Registry study methodology. BMC Res Notes. 2016;9(1):457.
27. Shih AW, Apelseth TO, Cardigan R, et al. Not all red cell concentrate units are equivalent: international survey of processing and in vitro quality data. Vox Sang. 2019;114(8):783–794.
28. Bellamy MC, Thomas D. Chapter 52 - Blood constituents and transfusion. In: Hemmings HC Jr, Hopkins PM, eds. Foundations of Anesthesia. 2nd ed. Edinburgh: Mosby; 2006:627–634.
29. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373–383.
30. McQuilten ZK, Zatta AJ, Andrianopoulos N, et al. Evaluation of clinical coding data to determine causes of critical bleeding in patients receiving massive transfusion: a bi-national, multicentre, cross-sectional study. Transfus Med. 2017;27(2):114–121.
31. van de Watering L. Pitfalls in the current published observational literature on the effects of red blood cell storage. Transfusion. 2011;51(8):1847–1854.
32. Trivella M, Stanworth SJ, Brunskill S, Dutton P, Altman DG. Can we be certain that storage duration of transfused red blood cells does not affect patient outcomes?BMJ. 2019;365:l2320.
33. Kanias T, Lanteri MC, Page GP, et al. Ethnicity, sex, and age are determinants of red blood cell storage and stress hemolysis: results of the REDS-III RBC-Omics study. Blood Adv. 2017;1(15):1132–1141.
Keywords:

Blood transfusion; massive transfusion; age of blood; storage time; red blood cells

Supplemental Digital Content

Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved.