Introduction

Approximately 112.5 million blood donations are annually collected worldwide, and the number of transfused red cells is 19.5 and 13.5 million units in China and the US, respectively.[1] Red cells are used to treat anemia; however, they are associated with side effects.[2] A decision regarding red-cell transfusion should be made when the benefits outweigh the risks and conserving red cell supplies and cost-effectiveness are also vital. Weighing the benefits and risks of transfusion triggered many clinical trials,[3–10] reviews,[11,12] and meta-analyses [13–17] to compare the liberal strategy with the restrictive strategy; however, debate still exists. The trigger and target hemoglobin levels for red-cell transfusion remain controversial, mainly owing to the diversity among patients. In a given clinical setting, a hemoglobin-based restrictive strategy may lead to oxygen supply/consumption imbalance, while a liberal strategy may result in exposure to unnecessary transfusion risk and exacerbation of blood shortage. Moreover, guidelines[18–20] and consensus[21] state the importance of an individualized approach and emphasize that a patient's medical status should be evaluated before making a transfusion decision. However, no guidelines suggest a practical way of objectively evaluating a patient's oxygen supply/consumption balance, and an individualized strategy has not been studied. Therefore, the optimal trigger and target hemoglobin levels for red-cell transfusion should be adjusted according to the factors, including but not limited to hemoglobin, as well as the other three equal-weighted factors in each patient. We proposed a score for individualized red-cell transfusion and conducted a pilot study in spine surgery. The results suggested that an objective evaluation of the oxygen supply/consumption balance and an individualized strategy with red-cell transfusion trigger and target could be achieved by application of the score and that the individualized strategy reduced the red cell requirement by 70% compared to that associated with a liberal strategy, without an increase in adverse events or prolonged length of hospital stay.[22] Then, we modified the previous score as the West-China-Liu's Score consisting of four factors (pulmonary function, cardiac function, and total body oxygen consumption, as well as the history of angina) for evaluating oxygen supply/consumption balance in total body and heart for individualized transfusion, and designed this trial to test the hypothesis that an individualized strategy can reduce the red cell requirement without compromising patient safety as compared with those associated with restrictive and liberal strategies in elective non-cardiac surgeries in adults.

Methods Ethical approval

The study protocol was approved by the Biological-Medical Ethical Committee of the West China Hospital of Sichuan University on January 10, 2012 (No. 2012(1)) and was registered at http://www.clinicaltrials.gov (NCT01597232). This open-label, multicenter, randomized clinical trial was conducted in 35 centers (Supplementary files, https://links.lww.com/CM9/B419) in China. An independent data and safety monitoring board approved the study protocol and monitored the study. Written informed consent was obtained from all participants.

Patients

Clinicians approached consecutive patients undergoing elective non-cardiac surgery with an estimated blood loss >1000 mL or 20% blood volume. Eligible patients were aged ≥14 years, and their habitation was <2500 m above sea level. Patients were ineligible if they declined to receive transfusions, the American Society of Anesthesiologists (ASA) classification was V or VI, or if they had severe hematological disorders or hemoglobinopathies, tumor metastasis, or multiple tumors.

Randomization and masking

The participants were randomized in a 1:1:1 ratio into three groups, stratified by the participating centers using the Statistical Analysis System (SAS) (Cary, NC, USA) Windows 9.1 programming. The group allocation was concealed in a non-transparent envelope with only a five-digit number of study IDs on the surface. Envelopes were distributed to each center where eligible patients were randomly assigned a study ID in the sequence according to the order of the patients recruited.

The investigator and care providers, but not the participants, research nurses for follow-up, or statisticians, were aware of the treatment allocation. All authors vouched for the data and analysis and the study's fidelity to the protocol.

Intervention

Both hemoglobin concentrations and the West-China-Liu's Score [Table 1] were simultaneously measured for all patients before and immediately after surgery, 24 h after randomization, on the discharge day, whenever red-cell transfusion was considered, and after each transfusion during the study.

Table 1 - West-China-Liu's Score. Points added Minimum FiO2 to keep SpO2 ≥95% (%) Adrenaline infusion rate (μg·kg−1·min−1) Core body temperature∗ (°C) History of angina 0 ≤35 Not required <38 No +1 36–50 ≤0.05 38–40 On exertion +2 ≥51 >0.05 >40 During normal daily living or at rest

∗Core body temperature is measured at either nasopharynx, oropharynx, tympanic membrane, rectum, or esophagus. Axillary temperature with 0.5°C added is accepted as core temperature. The West-China-Liu's Score consists of four items: (1) Minimum inhaled oxygen concentration to keep pulse oxygen saturation ≥95%, a clinical indicator for pulmonary function because it is not safe to measure the lowest pulse oxygen saturation while inhaling air in patients with poor pulmonary function. (2) Infusing rate of adrenalin required to maintain an adequate cardiac output. Here, adrenalin could be replaced by other inotropics with equivalent potent infusing rates because different institutions or medical groups may use different drugs as the first line inotropic. (3) Core body temperature, a clinical indicator of total body oxygen consumption. (4) History of angina and its severity. This is integrated into the score because the heart, an organ extracting and consuming most oxygen from blood, is the most sensitive to the oxygen supply/consumption misbalance in the body. The West-China-Liu's Score consists of 6 basal points and four items with maximum added 2 points for each item. The final score is the sum of 6 basal points plus all added points. If the sum point is ≥10, the final score will be counted as 10. The final West-China-Liu's Score ranges from minimum 6 up to maximum 10. These five levels of the score match with the five levels of hemoglobin concentration from lowest 6 g/dL up to highest 10 g/dL, which grade individual patients for their red-cell transfusion trigger and target at that time. If the final score is equal to or smaller than the instant hemoglobin concentration, red cells will not be transfused; if the score is higher than the value of the instant hemoglobin concentration, red cells will be transfused and the units of red cells required is twice the difference of the score minus hemoglobin concentration, because one unit of red cells is collected from 200 mL whole blood in China and the hemoglobin level could be increased by approximately 0.5 g/dL in most adults.

If a patient's hemoglobin concentration dropped to <10.0 g/dL during or after surgery, the patient was randomly assigned to one of the following three groups. In the restrictive group, red cells were transfused according to Chinese transfusion guidelines,[19] which state that red cells are generally required with a hemoglobin concentrations <7 g/dL and not required with a hemoglobin concentration ≥10 g/dL. When the hemoglobin concentration is between 7 and 10 g/dL, red-cell transfusion decisions should be based on physicians’ judgment regarding patients’ cardiopulmonary reserve, oxygen consumption, and age. In the liberal group, the red-cell transfusion trigger was hemoglobin concentration <9.5 g/dL and target was to maintain hemoglobin concentration >10 g/dL. In the individualized group, red-cell transfusion was guided by the West-China-Liu's Score [Table 1], which comprises six basal points and four items with a maximum of two points added for each. The final score is the sum of six basal points plus all added points (if the sum is greater than 10, the final score is counted as 10). Therefore, the final score ranges from 6 to 10. These five levels of the score match the five levels of hemoglobin concentration from the lowest 6 g/dL up to the highest 10 g/dL as both the red-cell transfusion trigger and target. If the final score is equal to or smaller than the current hemoglobin concentration, red cells will not be transfused; if the score is higher than the current hemoglobin concentration, red cell will be transfused and the units of red cells required are twice the difference of the score minus hemoglobin concentration because one unit of red cells is collected from 200 mL whole blood in China and elevates hemoglobin concentration by approximately 0.5 g/dL in most adults. Red blood cells were transfused within 30 min of the transfusion decision being made.

Research nurses telephoned the patients or legal guardians for follow-up. All data were collected via a web-based system, through which we monitored the participating centers for protocol adherence. Non-adherence to the protocol included red-cell transfusion without evaluation by the West-China-Liu's Score or red cells were transfused when hemoglobin level was >10 g/dL in all three groups or red cells were not transfused after the hemoglobin level was <7 g/dL in the restrictive group or <9.5 g/dL in the liberal group. Non-adherence data were included in the final analysis.

The principal investigator, study coordinator, and Office of Scientific Research at the West China Hospital were jointly responsible for all aspects of the study protocol and amendments. Ren Liao implemented the site monitoring. Three research nurses performed the data collection and follow-up.

Outcome measures

Two primary outcomes were evaluated in this study. First, the proportion of patients who received red blood cells (superiority test); and second, a composite of in-hospital complications (Supplementary Table 1, https://links.lww.com/CM9/B419) and all-cause mortality by day 30 (non-inferiority test).

The secondary outcomes included allogeneic blood cost and total in-hospital cost; all-cause mortality by day 60, day 180, and 1 year; in-hospital complications and infection rate; intensive care unit admission rate; incision healing status; and length of hospital stay after surgery.

Statistical analysis

The primary comparisons were between individualized, liberal and restrictive groups. In our pilot study,[22] 36.5% of the patients in the individualized group received red-cell transfusion, and we conservatively assumed a 20.0% difference between the two groups (36.5% vs. 56.5%) and then planned to recruit 1200 participants (400 in each group) (superiority test) for red-cell transfusion rate as the first primary outcome, and 4200 patients to allow for 3% of the non-inferiority margin of the composite rate between the individualized and restrictive groups as the second primary outcome with a difference of 2% (4% vs. 6%) within groups to be clinically important to achieve >99% power with an alpha level of 0.025 and an estimated 20% dropout rate. In December 2015, according to the original protocol plan, we reevaluated the sample size after 1200 patients were enrolled and found that 7659 patients would be needed for a non-inferiority margin of 3% points between the individualized and restrictive groups. The data and safety monitoring board approved a reduction in the recruitment of 1200 patients. This change resulted in an absolute change of approximately 4.5% in the non-inferiority margin of the composite rate.

All analyses followed the intention-to-treat principle and were performed using SAS 9.4 (SAS Institute, Cary, NC). Continuous variables were presented as means ± standard deviations or medians (interquartile ranges), as appropriate. Categorical variables were presented as n (percentage). We calculated the risk difference and odds ratio and used the Chi-square test to compare the primary outcomes among the three groups. The significance level was set at 2.5% to control for the type I error rate for the two primary comparisons. We also compared the proportions between the individual and each of the other two groups using Z tests. For continuous secondary outcomes, analysis of variance (ANOVA) or Kruskal–Wallis tests (if the normal distribution assumption was violated) were used for comparisons among the three groups. Group comparisons were performed using independent two-sample t tests or Wilcoxon rank-sum tests as appropriate. For discrete outcomes, Chi-square tests and odds ratios were used. Survival probabilities were calculated using Kaplan–Meier estimators and the comparisons were performed using methods by Klein et al.[23] For the hemoglobin level and the West-China-Liu's Score, we used a repeated-measurement ANOVA model with a random intercept for within-patient correlation. Two-sided P values <0.025 were considered statistically significant.

A sensitivity analysis was performed for the primary outcomes. We first fitted a random effect logistic model with transfusion requirement (or the composite of in-hospital complications and all-cause mortality by day 30) as the outcome and the group as the predictor, adjusting for age, sex, and ASA classification. The group effects were similar to those of the primary analysis in terms of showing the same direction and significance of the estimators. A random intercept was included to model correlations within each study center.

For the primary outcomes, we performed subgroup analysis by subgroups defined by age, ASA classification, and West-China-Liu's Score (scores of all peri-operative measurements were 6 or >6 at least once). Forest plots were drawn based on odds ratios and corresponding 95% confidence intervals (CIs).

Results

From May 17, 2012, to January 18, 2016, 1351 eligible patients signed the consent, of whom 80 had tumor metastasis, 1 withdrew consent, and 57 had hemoglobin concentrations not dropping <10.0 g/dL. A total of 1213 patients underwent randomization, and three patients were withdrawn at the physician's request. A total of 1210 patients received the assigned intervention, 26 of whom were withdrawn at the physician's request and two at the patient's request. A total of 1182 patients were included in the final analysis as follows: 379, 419, and 384 in the individualized, restrictive, and liberal strategies, respectively [Figure 1]. A telephonic follow-up was performed on January 20, 2017. The patient's baseline characteristics were similar among the three groups [Table 2].

Figure 1:

CONSORT flow diagram of red-cell transfusion strategies for non-cardiac surgery in adults.

Table 2 - Comparison of the baseline characteristics and surgical types among the three groups with different red-cell transfusion strategy. Characteristics Individualized group (n = 379) Restrictive group (n = 419) Liberal group (n = 384) Age (years) 48.9 ± 15.6 50.4 ± 15.1 51.6 ± 15.4 Male 119 (31.4) 151 (36.0) 152 (39.6) Weight (kg) 60.0 ± 10.8 59.8 ± 11.2 60.3 ± 11.7 Body mass index (kg/m2) 23.0 ± 3.5 22.7 ± 3.7 22.8 ± 3.8 ASA physical status  I or II 317 (83.6) 357 (85.2) 308 (80.2)  III or IV 62 (16.4) 62 (14.8) 76 (19.8) Cardiovascular disease 29 (7.7) 34 (8.1) 33 (8.6) Surgery type  General 129 (34.0) 148 (35.3) 134 (34.9)  Orthopedic 139 (36.7) 158 (37.7) 153 (39.8)  Thoracic 7 (1.9) 6 (1.4) 7 (1.8)  Neurosurgical 30 (7.9) 36 (8.6) 38 (9.9)  Urological 11 (2.9) 15 (3.6) 9 (2.3)  Gynecological 59 (15.6) 54 (12.9) 42 (10.9)  Others 4 (1.1) 2 (0.5) 1 (0.3) Tumor 133 (35.1) 157 (37.5) 151 (39.3)

Data are shown as n (%) or mean ± standard deviation. ASA: American Society of Anesthesiologists. Data from the full analysis set population are presented here. In the individualized group, red-cell transfusion trigger and target were determined according to the West-China-Liu's Score. In the restrictive group, decision of red-cell transfusion was made in accordance with the current Chinese transfusion guidelines, which state that red-cell transfusion is usually required with the Hb level <7 g/dL; and usually not required with the hemoglobin level ≥10 g/dL; when the hemoglobin concentration is between 7 g/dL and 10 g/dL, decision of red-cell transfusion should be based on the physician's judgment about patients’ condition. In the liberal group, the red-cell transfusion trigger was a Hb level <9.5 g/dL and red-cell transfusion target was to maintain Hb level ≥10 g/dL. No statistically significant differences were found among the three groups for any of the listed variables.

Primary outcomes

Of the patients receiving the individualized strategy, 30.6% (116/379) received red-cell transfusion, which was significantly less than the proportions in the restrictive (62.5% [262/419]; absolute risk difference: 31.92% [97.5% CI: 24.42–39.42%]; P<0.001) and liberal strategies (89.8% [345/384]; absolute risk difference: 59.24% [97.5% CI: 52.91%–65.57%]; P<0.001; Table 3). The composite of in-hospital complications and all-cause mortality by day 30 did not statistically significantly differ among the three groups (P = 0.224; Table 3). The composites were similar between the individualized and restrictive groups (11.6% [44/379] vs. 11.5% [48/419]; absolute risk difference: −0.15% [97.5% CI: −5.23% to 4.92%]; P = 0.946), and similar between the individualized and liberal groups (11.6% [44/379] vs. 15.1% [58/384]; absolute risk difference: 3.49% [97.5% CI: −2.02% to 9.01%]; P = 0.156). For the non-inferiority test, we set the non-inferiority margin to 4.5%. The individualized group was non-inferior to the liberal group, as the lower bound (−2.02%) was greater than −4.50%. The individualized group was not significantly non-inferior to the restrictive group, as the lower bound (−5.23%) of the CI was smaller than −4.50%. However, the risk difference was −0.15%, which was close to 0, and no statistically significant difference between the risks of the two groups existed.

Table 3 - Primary and secondary outcomes among the three groups with different red-cell transfusion strategy. Outcomes Individualized group (n = 379) Restrictive group (n = 419) Liberal group (n = 384) Odds ratio (97.5% CI) Risk difference (97.5% CI) Mean difference (95% CI) P value Primary outcomes  Patients received red cell transfusion 116 (30.6) 262 (62.5) 345 (89.8) <0.001 3.78 (2.70–5.30)† <0.001† 20.06 (12.74–31.57)‡ <0.001‡  Composite of in-hospital complications and mortality by day 30 44 (11.6) 48 (11.5) 58 (15.1) 0.224 0.99 (0.60–1.62)† −0.15 (−5.23, 4.92) † 0.946† 1.35 (0.84–2.19)‡ 3.49 (−2.02, 9.01)‡ 0.156‡ Secondary outcomes  Units of red cell transfused per person∗ 0 (0–2.0)¶ 2.0 (0–4.0)¶ 3.5 (2.0–4.5)∗∗ <0.001 1.57 (0.90–2.23)† <0.001† 2.75 (2.07–3.42)‡ <0.001‡  Patients received FFP transfusion 84 (22.1)¶ 174 (41.7)¶ 175 (45.8)∗∗ <0.001  Patients received platelet transfusion 13 (3.4) 4 (1.0)¶ 9 (2.4)∗∗ 0.059  Patients received cryoprecipitate transfusion 11 (2.9) 8 (1.9)¶ 11 (2.9)∗∗ 0.598  Patients received albumin transfusion 60 (15.8) 86 (20.6)¶ 75 (19.6)∗∗ 0.195  Hemoglobin concentration (g/dL)   Preoperation 10.8 ± 1.7 10.8 ± 1.7 11.2 ± 1.8 0.092   Completion of operation 8.6 ± 1.3 9.0 ± 1.3 10.4 ± 1.3 <0.001   24 h after operation 9.1 ± 1.6 9.6 ± 1.7 10.6 ± 1.5 <0.001   Before discharge 9.6 ± 1.4 9.8 ± 1.4 10.7 ± 1.3 <0.001   Before all transfusions 6.5 ± 1.6 7.7 ± 1.3 8.5 ± 1.0 <0.001  West-China-Liu's Score   Preoperation 6.1 ± 0.4 6.0 ± 0.3 6.1 ± 0.3 0.128   Completion of operation 6.2 ± 0.6 6.1 ± 0.4 6.1 ± 0.4 0.033   24 h after operation 6.3 ± 0.5 6.2 ± 0.5 6.2 ± 0.5 0.258   Before discharge 6.1 ± 0.4 6.0 ± 0.2 6.1 ± 0.3 0.128   Before all transfusions 7.0 ± 1.0 6.3 ± 0.7 6.2 ± 0.6 <0.001  Scores of all measurements were 6 240 (63.5) 283 (69.4) 268 (72.4) 0.028   Mortality by day30 4 (1.1) 2 (0.5) 8 (2.3) 0.116   Mortality by day60 4 (1.1) 3 (0.8) 14 (4.1) 0.013   Mortality by day180 12 (3.6) 12 (3.2) 21 (6.1) 0.126   Mortality by 1 year 22 (7.6) 25 (7.7) 33 (10.9) 0.062  In-hospital complications§ 44 (11.6) 48 (11.5) 57 (14.8) 0.273 0.98 (0.64–1.52)† 0.946† 1.33 (0.87–2.02)‡ 0.187‡   Cardiac 2 (0.5) 1 (0.2) 7 (1.8) 0.035   Central nervous 4 (1.1) 4 (1.0) 5 (1.3) 0.890   Pulmonary 13 (3.4) 19 (4.5) 19 (5.0) 0.565   Digestive 9 (2.4) 6 (1.4) 8 (2.1) 0.611   Urinary/reproductive 3 (0.8) 4 (1.0) 5 (1.3) 0.771   Post-operative bleeding 1 (0.3) 2 (0.5) 1 (0.3) 0.830   Others 22 (5.8) 25 (6.0) 31 (8.1) 0.365  In-hospital infection 18 (4.8) 26 (6.2) 29 (7.6) 0.274  ICU admission rate 68 (17.9) 77 (18.5) 83 (21.7) 0.354  Healing status of surgical incision 0.636   I 372 (96.1) 396 (94.5) 360 (93.8)   II 11 (2.8) 18 (4.2) 20 (5.2)   III 4 (1.0) 5 (1.3) 4 (1.0)  Stitch removal time (days) 15.5 ± 6.7 15.8 ± 8.9 14.2 ± 13.3 0.280  Length of hospital stay (days) 21.5 ± 13.9 21.6 ± 13.1 21.8 ± 14.2 0.796  Length of hospital stay after surgery (days) 12.4 ± 11.3 12.2 ± 9.9 12.3 ± 9.9 0.732  Transfusion related cost   (Thousand ¥ per person) 0.55 ± 1.4 1.06 ± 1.6 1.52 ± 2.2 <0.001 0.51 (0.29, 0.72)† <0.001† 0.97 (0.71, 1.24)‡ <0.001‡  Total in-hospital cost   (Thousan