Key Points
Question
What are the current practice patterns of acute normovolemic hemodilution (ANH) use in adult cardiac surgery?
Findings
In this cohort study of 16 795 patients who underwent coronary artery bypass grafting and/or valve replacement with cardiopulmonary bypass, ANH utilization was low (14.7%) in the US. In propensity score-matched cohorts, ANH was associated with a 27% reduction in overall transfusion odds, with the effect primarily based on larger-volume ANH collections.
Meaning
Results of this study suggest that ANH may have the potential to improve blood supply and resource utilization for the health care system.
Abstract
Importance
Acute normovolemic hemodilution (ANH) is a well-known blood conservation technique, yet its adoption in the US remains low (<20%), and cardiac surgery remains the largest consumer of blood components. Increasing vulnerabilities in the US blood supply underscore the need to reassess ANH effectiveness in blood conservation.
Objective
To evaluate ANH use in perioperative transfusion and blood component use in patients undergoing surgery with cardiopulmonary bypass (CPB).
Design, Setting, and Participants
A retrospective propensity–score matched cohort study using data from the Society of Thoracic Surgeons (STS) Adult Cardiac Surgery Database (version 4.20.2). The study included adult patients (≥18 years) undergoing coronary artery bypass grafting (CABG) and/or valve surgery with CPB (July 2020 to September 2023).
Exposure
The primary exposure was ANH, defined as any recorded use/volume in the STS database.
Main Outcome and Measures
The primary outcome was intraoperative or postoperative transfusion of any blood component. The volume-dependent ANH effect estimate was also assessed using a cutoff value of 650 mL. Secondary outcomes included the total number of transfused red blood cells (RBC) and non-RBC components (platelets, plasma, and cryoprecipitate); bleeding reexplorations; length of stay in the intensive care unit; and estimates for blood acquisition and activity-based costs.
Results
Among 16 795 patients (mean [SD] age, 65.3 [10.5] years; 12 114 male [72.1%]), ANH was reported in 2463 cases (14.7%). Patients receiving ANH had a significantly lower preoperative anemia rate and higher baseline hematocrit (mean difference: 2.5%; 95% CI, 2.3-2.7; P < .001). After propensity score matching (n = 2282 pairs), transfusion rates were significantly lower in the ANH group (31.2% vs 36.4%; P < .001). ANH was associated with 27% lower odds of any transfusion (odds ratio, 0.73; 95% CI, 0.60-0.89). High-volume ANH (≥650 mL) further reduced the odds of transfusion by 47% to 64% for both RBC and non-RBC components. The cumulative number of transfused erythrocyte and platelet units was 167 and 295 units lower, respectively, in the ANH group compared with the control group, with lower estimates for both acquisition and activity-based costs.
Conclusions and Relevance
In this study, a volume-dependent association was found between ANH and reduced RBC and non-RBC transfusion rates in patients undergoing cardiac surgery. Despite being safe and cost-effective, ANH remains underused as a blood conservation strategy.
This cohort study evaluates acute normovolemic hemodilution in perioperative transfusion and blood component utilization in cardiac surgery patients undergoing surgery with cardiopulmonary bypass.
Introduction
Global demand for cardiac surgery is increasing, with more than 1 million procedures performed annually worldwide.1 In high-income countries, where the average surgical rate is 123.2 cases per 100 000 population,1 cardiac surgery remains the largest consumer of blood components, with 30% to 50% of patients receiving red blood cell (RBC) transfusions.2,3,4 While a steady supply of allogeneic blood components is essential, the recent COVID-19 pandemic and natural disasters have revealed vulnerabilities in the US blood supply.5,6
Acute normovolemic hemodilution (ANH) is a blood conservation strategy in which a patient’s whole blood is withdrawn at the start of surgery and subsequently replaced with crystalloid or colloid solutions to maintain normovolemia. In cardiac surgery, the collected autologous blood precludes not only RBCs, but also platelets and coagulation factors, from exposure to cardiopulmonary bypass (CPB). Thus, when reinfused after protamine administration, ANH blood not only provides RBCs to correct anemia but also supports hemostasis through the delivery of platelets and coagulation factors.7 This technique has been recognized as a valid blood conservation technique since the 1970s.8 Although recommended in patient blood management guidelines and consensus statements,9,10 its adoption in the US remains low (<20%).11 Limited implementation of high-volume ANH (≥650 mL in collected blood volume) may be attributed to mixed evidence and the relatively modest RBC reduction observed (−0.79 units; 95% CI, −1.25 to −0.34 units; P = .001) in a meta-analysis.8 However, even a small reduction in blood usage could significantly impact overall demand, particularly given the rising cost of blood components12 and the more than 300 000 adult cardiac surgical cases performed annually in the US.
Beyond hemoglobin preservation, recent retrospective studies have reported an association between ANH and reductions in RBC and non-RBC components (plasma, platelets, and cryoprecipitate).7,13,14,15 These single-center studies all used large-volume blood collection during ANH, confirming a dose-dependent hemostatic effect estimate, as previously observed in the 2015 STS database analysis.11 Validating the efficacy of ANH in RBC conservation and its potential hemostatic benefits is critical to understanding its utility in a broader and more heterogeneous patient population in modern cardiac surgical practice.
Our primary study aim was to evaluate current practice patterns of ANH use in the US. A secondary aim was to determine whether the outcomes of ANH use in these end points remains consistent across Society of Thoracic Surgeons (STS) patient populations undergoing coronary artery bypass grafting (CABG) and/or valve repair/replacement, stratified by STS risk score and CPB duration. We hypothesized that ANH would be associated with reductions in perioperative RBC and non-RBC transfusion requirements and might influence rates of reexploration for bleeding and other perioperative adverse events.
Methods
Study Population
A retrospective cohort study was performed using data from the STS Adult Cardiac Surgery database version 4.20.2 and included patients who underwent CABG and/or valve replacement (July 2020 to September 2023) with CPB. The STS database, including the anesthesia module, was evaluated to extract data on blood conservation strategies, transfusion practices, and related outcomes reported by institutions across the US. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline and was reviewed and approved as research exempt from need for informed consent by the STS research review committee and the University of Oklahoma institutional review board.
Exposure
Use of ANH was the primary exposure of interest defined dichotomously as ANH performed vs ANH not-performed, as recorded in the STS database. For this definition, any volume of ANH was considered as use of ANH. The ratio of ANH volume to fluid replacement, which varies between 1:1 to 1:3 was unavailable in the database. For a secondary analysis, the volume-dependent effect of ANH was explored using a 3-level variable based on a cutoff value from a previous meta-analysis8; ANH less than 650 mL, ANH greater than or equal to 650 mL, and no ANH (control).
Outcomes and Covariates
The primary outcome of interest was any intraoperative or postoperative blood component transfusion. Secondary outcomes included the total number of units (U) of all RBC and non-RBC components (platelets, plasma, and cryoprecipitate), reexploration for bleeding, and the length of intensive care unit (ICU) stay.
Covariates were chosen according to clinical relevance as determined by 2 cardiac anesthesiologists (K.T. and M.M.) with expertise in the use of ANH. Selected variables included age; sex; body mass index (BMI); hypertension; left ventricular ejection fraction; preoperative hematocrit; anemia (hemoglobin <13 mg/dL for both sexes16 [to convert hemoglobin to g/L, multiply by 10]); diabetes; liver disease, baseline creatinine (mg/dL); platelets defined as less than 150 or greater than or equal to 150 (×103/µL [to convert platelet count to ×109/L, multiply by 1); aspirin or P2Y12 inhibitor use within 5 days; type of surgery (CABG and/or valve repair/replacement); retrograde autologous priming (RAP); total CPB time. BMI groups were formed as 16 to 24, 25 to 30, 31 to 40, and greater than 40 (calculated as weight in kilograms divided by height in meters squared).
Post Hoc Analysis of ANH in Blood Component Costs
Blood component acquisition and activity-based costs for 2024 were estimated using inflation-adjusted values from existing literature.12,17,18,19 Activity-based costs account not only for acquisition costs but also for indirect in-hospital process costs and overhead expenses.17 The estimated costs (acquisition cost; activity-based cost) were as follows: RBC ($334; $1024), platelets ($762; $1771), plasma ($51; $497), and cryoprecipitate ($526; $1221). The consumable costs for ANH, including citrate bags and tubing, were approximately $10 per bag and were considered negligible. In the matched cohort, the costs for RBCs, platelets, plasma, and cryoprecipitate were calculated by multiplying the number of units and per-unit acquisition or activity-based cost.
Statistical Analysis
Patient variables and outcomes were summarized for the overall group and by ANH use. Associations between categorical variables and ANH group were tested using the χ2 test. Differences between continuous variables were tested using t tests. Missing data among variables used for the propensity score model and primary analysis were minimal with the highest proportion being 1.2% for left ventricular ejection fraction; therefore, imputation was not performed. The aforementioned variables, with the exception of CPB time, were used in a logistic regression model to develop a propensity score for ANH use (eTable 1 in Supplement 1). An exact match was used for sex, BMI group, and surgery type. The propensity score model did not include CPB time, as it is not known prior to ANH, the intervention of interest. Patients receiving ANH and patients not receiving ANH were matched on the log of the propensity score (caliper = 0.149) using a built-in greedy nearest neighbor algorithm20 via the PSMatch procedure using SAS software version 9.4 (SAS Institute). Standardized mean differences (SMDs) were calculated to assess covariate balance, with an SMD less than 0.1 indicating adequate balance. Statistical significance was set at 2-sided P < .05.
Outcome variables were summarized for the matched groups. A mixed logistic regression model was used to determine the effect of ANH on the odds of overall transfusion and separately for each blood component. The models included ANH or 3-level ANH and total perfusion time as fixed effects and institution (reporting site) as a random effect. Models were also evaluated with the propensity score match ID variable as a second random effect to incorporate the 1:1 match.
Results
Among a total of 16 795 individuals from 52 sites (mean [SD] age, 65.3 [10.5] years; 12 114 male [72.1%]; 4681 female [27.9%]), ANH was performed at 28 sites, reporting 2463 cases (14.7%) (Table 1; eFigure in Supplement 1). Although overall, differences in covariates between ANH and control groups were not large, there were fewer women in the ANH group (21.8% vs 28.9%; P < .001). The baseline hematocrit was significantly higher in the ANH group, with a mean difference of 2.5% (95% CI, 2.3%-2.7%). Rates of preoperative anemia in the ANH group were one-half of those in the control group. Of 16 795 individuals, transfusion of some blood component was documented in 6966 patients (41.5%). The number of documented transfusions was lower in the ANH group for both RBC (21.3% vs 33.2%; P < .001) and non-RBC components (21.3% vs 33.2%; P < .001) compared with the non-ANH group.
Table 1. Demographic and Clinical Characteristics Before Propensity Score Matching Overall and by ANH Status.
| Variable | Participants, No. (%) | ||
|---|---|---|---|
| Overall (N = 16 795) | ANH (n = 2463) | Control (n = 14 332) | |
| Age, mean (SD), y | 65.3 (10.5) | 64.7 (10.1) | 65.4 (10.5) |
| Sex | |||
| Female | 4681 (27.9) | 537 (21.8) | 4144 (28.9) |
| Male | 12 114 (72.1) | 1926 (78.2) | 10 188 (71.1) |
| Surgery type | |||
| CABG only | 8546 (50.9) | 1158 (47.0) | 7388 (51.6) |
| Valve replacement or repair only | 6065 (36.1) | 1016 (41.3) | 5049 (35.2) |
| Combined CABG and valve replacement or repair | 2184 (13.0) | 289 (11.7) | 1895 (13.2) |
| BMI, mean (SD) | 29.4 (6.4) | 29.5 (6.7) | 29.4 (6.4) |
| BMI group | |||
| 16-24 | 3392 (20.2) | 459 (18.6) | 2933 (20.5) |
| 25-30 | 7171 (42.7) | 1097 (44.6) | 6074 (42.4) |
| 31-39 | 5419 (32.3) | 782 (31.8) | 4637 (32.4) |
| ≥40 | 795 (4.7) | 124 (5.0) | 671 (4.7) |
| Hypertension | 14 158 (84.3) | 1873 (76.1) | 12 285 (85.7) |
| Diabetes | 6536 (38.9) | 848 (34.4) | 5688 (39.7) |
| Liver disease | 539 (3.2) | 86 (3.5) | 453 (3.2) |
| P2Y12 inhibitor use | 539 (3.2) | 54 (2.2) | 485 (3.4) |
| Aspirin use | 9844 (58.6) | 1370 (55.6) | 8474 (59.1) |
| Ejection fraction, mean (SD) | 55.8 (10.5) | 56.8 (10.0) | 55.7 (10.6) |
| Anemia | 5026 (30.1) | 383 (15.6) | 4643 (32.5) |
| Hematocrit, mean (SD), % | 41.3 (5.0) | 43.4 (4.4) | 41.0 (5.0) |
| Creatinine, mean (SD), mg/dL | 1.2 (1.2) | 1.1 (0.7) | 1.2 (1.3) |
| Platelets <150 × 103/µL | 1481 (8.9) | 188 (7.7) | 1293 (9.1) |
| RAP use | 10 399 (62.4) | 1606 (65.6) | 8793 (61.8) |
| Antifibrinolytic | 15 521 (95.6) | 2389 (97.0) | 13 132 (91.6) |
| CPB duration, mean (SD), min | 122.5 (45.2) | 124.9 (44.4) | 122.1 (45.3) |
| Any blood component | 6966 (41.5) | 774 (31.4) | 6192 (43.2) |
| Any RBC | 5284 (31.5) | 524 (21.3) | 4760 (33.2) |
| Any non-RBC component | 4423 (26.3) | 512 (20.8) | 3911 (27.3) |
| Any plasma | 2581 (15.4) | 306 (12.4) | 2275 (15.9) |
| Any cryoprecipitate | 2397 (14.3) | 267 (10.8) | 2130 (14.9) |
| Any platelets | 2988 (17.8) | 330 (13.4) | 2658 (18.6 |
| Intensive care unit, mean (SD), h | 68.2 (79.9) | 66.1 (74.7) | 68.6 (80.8) |
| Reoperation for bleeding or tamponade | 401 (2.4) | 53 (2.2) | 348 (2.4) |
Abbreviations: ANH, acute normovolemic hemodilution; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); CABG, coronary artery bypass grafting; CPB, cardiopulmonary bypass; P2Y12 inhibitor, clopidogrel, prasugrel, or ticagrelor; RAP, retrograde autologous priming; non-RBC component, platelet concentrate, plasma, or cryoprecipitate; RBC, red blood cells.
SI conversion factors: To convert hematocrit to proportion of 1.0, multiply by 0.01; platelet count to ×109/L, multiply by 1; creatinine to micromoles per liter, multiply by 88.4.
The matching procedure resulted in 2282 pairs matched based on the specified parameters (Figure; eTable 1 in Supplement 1). All variables showed adequate balance, as indicated by SMD values (eTable 2 in Supplement 1). The mean (SD) CPB durations were 124.7 (44.1) minutes in the ANH group and 121.3 (45.1) minutes in the control group. The rates of retrograde autologous priming (RAP) use were similar between the groups. The proportion of patients receiving any blood component transfusion was lower for ANH (31.2%) compared with the control group (36.4%) (Table 2). Lower proportions of transfusion were also noted among patients receiving ANH for RBC and non-RBC components. The mean number of reported units transfused per cohort was not statistically different, except for lower mean platelet transfusion in the ANH group. There were no differences between the 2 groups in reoperation rates for bleeding or tamponade, and ICU length of stay was comparable (Table 2). The addition of match ID as a second random effect resulted in only negligible differences in model estimates; therefore, results were presented from models using only institution ID as a random effect.
Figure. Standardized Mean Difference Between Acute Normovolemic Hemodilution (ANH) and Control Groups Prematching and Postmatching for Demographic and Clinical Characteristics.
Standardized difference between the group with ANH and the other without ANH are shown for demographic and clinical characteristics before and after propensity score matching.
Table 2. Summary of Outcomes for the Propensity Score–Matched Cohorts.
| Outcome | ANH (n = 2282) | No ANH (n = 2282) | P value |
|---|---|---|---|
| Blood component transfusions | |||
| Any blood component, No. (%) | 712 (31.2) | 830 (36.4) | <.001 |
| RBC, No. (%) | 480 (21.0) | 547 (24.0) | .02 |
| Platelets, No. (%) | 308 (13.5) | 352 (15.4) | .06 |
| Plasma, No. (%) | 290 (12.7) | 310 (13.6) | .38 |
| Cryoprecipitate, No. (%) | 248 (10.9) | 331 (14.5) | <.001 |
| Any blood component, mean (SD), units | 1.40 (3.56) | 1.62 (4.02) | .05 |
| RBC, mean (SD), units | 0.54 (1.47) | 0.62 (2.00) | .16 |
| Platelets, mean (SD), units | 0.28 (0.87) | 0.41 (1.57) | .001 |
| Plasma, mean (SD), units | 0.31 (0.97) | 0.31 (0.98) | .93 |
| Cryoprecipitate, mean (SD), pools | 0.28 (1.42) | 0.29 (0.85) | .69 |
| Secondary end points | |||
| Intensive care unit, mean (SD), h | 65.6 (75.5) | 64.3 (73.2) | .54 |
| Reoperation for bleeding, No. (%) | 50 (2.2) | 50 (2.2) | .41 |
Abbreviations: ANH, acute normovolemic hemodilution; RBC, red blood cells.
Compared with the control group, ANH use was associated with 27% lower odds for any blood component exposure during the index hospitalization (odds ratio [OR], 0.73; 95% CI, 0.60-0.89) (Table 3). Model details are available in in eTable 3 in Supplement 1. Estimates for specific components varied, but odds remained significantly lower for patients receiving ANH with respect to transfusions of any RBC (adjusted OR, 0.67; 95% CI, 0.55-0.83; P < .001) and any platelet (adjusted OR, 0.71; 95% CI, 0.55-0.92; P = .009). In the secondary analysis, low (<650 mL) vs high (≥650 mL) volumes of ANH were compared with the control for transfusion odds (Table 3). Model details are available in eTable 4 in Supplement 1. ANH volume was missing for 11% of patients receiving ANH. To use a conservative analytical approach, all patients with missing volume were assigned to the low-volume ANH group. Patients receiving high-volume ANH had much lower odds of transfusion compared with the control group for any blood component with a reduction in odds ranging from 47% for any plasma transfusion to 64% for RBC transfusion.
Table 3. Effects of ANH and Its Volume on Transfusion Requirements in the Matched Cohortsa.
| Transfusion requirement | ANH, AOR (95% CI) | ||
|---|---|---|---|
| Any volume | Volume ≥650 mL | Volume <650 mL | |
| Overall transfusion | 0.73 (0.60-0.89)b | 0.46 (0.36-0.59)b | 0.91 (0.74-1.12) |
| Any RBC | 0.67 (0.55-0.83)b | 0.36 (0.27-0.48)b | 0.88 (0.70-1.09) |
| Any platelet | 0.71 (0.55-0.92)b | 0.50 (0.36-0.70)b | 0.84 (0.64-1.11) |
| Any plasma | 0.90 (0.69-1.17) | 0.53 (0.37-0.76)b | 1.14 (0.86-1.51) |
| Any cryoprecipitate | 0.83 (0.63-1.08) | 0.52 (0.36-0.75)b | 1.02 (0.77-1.36) |
Abbreviations: ANH, acute normovolemic hemodilution; AOR, adjusted odds ratio; RBC, red blood cells.
All models were adjusted for perfusion time using the control group (no ANH) as the referent.
The CIs for estimates that do not include 1 are statistically significant at P < .05.
According to the number of transfused blood components per cohort, ANH had the greatest impact on RBC and platelet use: the cumulative number of transfused erythrocyte and platelet units was 167 and 295 units lower, respectively, in the ANH group compared with the control group, with lower estimates for both acquisition and activity-based costs (Table 4). Given the high acquisition and activity-based costs of RBC and platelet units, both costs were significantly reduced in the ANH group compared to the control group per 2282 patients. When normalized to 10 000 patients, cost reductions were estimated to be $1.2 million for acquisition and $3.1 million for activity-based costs.
Table 4. Acquisition and Activity-Based Cost Estimates of Blood Components for the Matched Cohorts.
| Blood componenta | Cohort | Differenceb | |
|---|---|---|---|
| ANH | Control | ||
| RBC | |||
| No. | 1241 | 1408 | −167 |
| Acquisition cost | $499 354 | $566 551 | $(67 197) |
| Activity-based cost | $1 531 270 | $1 737 331 | $(206 061) |
| Platelet | |||
| No. | 635 | 930 | −295 |
| Acquisition cost | $443 522 | $649 568 | $(206 046) |
| Activity-based cost | $1 031 157 | $1 510 199 | $(479 042) |
| Plasma | |||
| No. | 699 | 705 | −6 |
| Acquisition cost | $39 459 | $39 797 | $(339) |
| Activity-based cost | $385 373 | $388 681 | $(3308) |
| Cryoprecipitate | |||
| No. | 630 | 661 | −31 |
| Acquisition cost | $331 241 | $347 541 | $(16 299) |
| Activity-based cost | $768 984 | $806 823 | $(37 839) |
| Total cost difference (n = 2282 per group) | |||
| Acquisition cost | NA | NA | $(289 881) |
| Activity-based cost | NA | NA | $(726 249) |
Discussion
In this cohort study using propensity–score matched analysis of the STS database, ANH utilization was found to be low (14.7%) among adult patients undergoing CABG and/or valve procedures with CPB. Despite this finding, ANH was associated with a 27% reduction in overall transfusion odds, with the finding primarily based on larger-volume ANH collections (≥650 mL). High-volume ANH collection was associated with 47% to 64% lower odds for transfusion of various blood components. Although mean transfused units did not differ between the matched cohorts, cumulative transfusion acquisition and activity-based costs were significantly lower in the ANH group.
Our findings align with prior single-institution observations7,13 and propensity-matched analyses,14,15 supporting positive outcomes of ANH. In a small observational study (n = 68 per cohort), patients receiving ANH (median 1100 mL) had a significantly higher transfusion-free rates (75.0% vs 54.8%; P = .006).13 After risk adjustments for platelet count and the type of surgery, lower rate ratios were found for RBC transfusion (0.58; 95% CI, 0.39-0.88; P = .009) and for non-RBC transfusion (0.63; 95% CI, 0.44-0.89; P = .01). In high bleeding risk aortic surgery using deep hypothermia, Geube et al15 found ANH greater than or equal to 800 mL was associated with significantly reduced intraoperative RBC requirements (OR, 0.46; 95% CI, 0.24-0.88; P = .02), while the effects of lower volumes (<800 mL) were limited (OR, 0.60; 95% CI, 0.24-1.4; P = .20). In their study, the intraoperative benefit of ANH was offset by postoperative transfusions due to bleeding and transfusions in the ICU. In contrast, the present study found a sustained reduction for the entire hospital stay in the ANH cohort compared with the control group (Table 3). This discrepancy may be explained by differences in surgical complexity, particularly the lack of routine deep hypothermia and shorter CPB duration in the study cohort. Although the optimal ANH volume remains elusive,7,8,13,14,15 findings of the present study may support the benefit of a larger volume collection (≥650 mL). Finally, the results of a recent multinational prospective randomized trial of ANH vs usual care involving 2010 patients undergoing cardiac surgery did not find a significant benefit of ANH.21 This trial did not demonstrate ANH efficacy, reporting the RBC transfusion rates of 27.3% with ANH vs 29.2% (relative risk, 0.93; 95% CI, 0.81-1.07; P = .34). Notably, the median volume of ANH collected was 650 mL at the lower threshold of what prior meta-analyses have defined as high volume.8 The lower BMI of their study population (26), compared with this study (29.5), along with intercenter variability in transfusion protocols, may have limited their ability to detect a meaningful reduction in transfusion rates.21
The relevance of ANH in the presence of other blood conservation techniques warrants discussion. Retrograde autologous priming (RAP) reduces hemodilution by incorporating patient blood into CPB priming, and it is associated with a lower OR for intraoperative RBC usage (OR, 0.34; 95% CI, 0.22-0.55; P < .001).22 In our dataset, RAP was reported in over 60% of both matched cohorts. Even after having matched for RAP, the association of ANH with the lower odds for transfusion remained significant, supporting its additive effect in multimodal patient blood management.
To our knowledge, this study represents the largest retrospective analysis of ANH in commonly performed cardiac surgeries in the US. Despite its low adoption, propensity-matched analysis supports the role of ANH in transfusion reduction, with 27% lower odds for any transfusion. Critical shortage of platelet concentrates is not infrequent due to chronic donor shorage23; thus, our finding of 50% lower odds for platelet transfusion associated with high-volume ANH suggesting its potential as a means to mitigate vulnerability in the platelet supply chain. Although transfusion risk reduction per patient may be relatively modest, the cumulative impact is substantial, translating into millions per year in savings at larger scales in a patient population exceeding 10 000. The principle of aggregation of marginal gains also applies to ANH.24 Given the high US demand for cardiac surgery, estimated at 271.5 cases per 100 000 annually, and the increasing need for blood conservation strategies in resource-limited settings,1 ANH should be recognized in the US as a cost-effective method to optimize global blood resource utilization.
Limitations
The present study has limitations. First, as a retrospective observational study, there remains the possibility of residual confounding and incomplete risk adjustment. Variability in blood conservation practices, including preoperative iron therapy25 or exact hemoglobin thresholds for RBC transfusion was not fully accounted for, although baseline hematocrit was matched via propensity score. Institutions using multimodal patient blood management strategies may inherently demonstrate lower transfusion rates.26 The use of RAP did not affect ANH with lower odds for transfusion, but the outcomes associated with preoperative anemia management, which may affect both RBC transfusion rates and the feasibility of ANH, could not be ascertained. Additionally, we cannot rule out the possibility that other strategies not captured in the STS data may have been factors in the observed outcomes. Further, the generalizability of our findings may be limited by selection bias introduced through the exclusion of unmatched observations, as well as by the high proportion of patients with BMI greater than 30 in the STS database (37%), particularly to populations with lower average BMI outside the US21,27 Finally, the ANH volumes reported in these data lack sufficient granularity to evaluate potential optimal volumes or determine whether ANH was applied most effectively.
Conclusions
In this cohort study using propensity–score matched analysis of the STS adult cardiac surgery database, a dose-dependent association was found between the volume of ANH and reduced allogeneic RBC and non-RBC transfusion rates following CABG and valve surgeries with CPB. As a low-cost, easy-to-perform intervention, ANH could be incorporated alongside other conservation strategies, such as preoperative iron therapy and RAP, to minimize transfusion-related risks and to improve blood supply and resource utilization for the US health care system. Future research should explore combining ANH with factor concentrates28 to enhance its feasibility and hemostatic effectiveness. Given global variability in surgical populations and procedures, additional studies are needed to assess ANH’s broader applicability across diverse populations and health care settings.
eTable 1. Propensity Score Model Parameter Estimates
eTable 2. Post-Match Variables in the Propensity Score Model
eTable 3. Parameter Estimates for Model of ANH (Yes/No) for Any Blood Component Transfusion
eTable 4. Parameter Estimates for Model of ANH volume categories (3-level) for Any Blood Component Transfusion
eFigure 1. Study Population and Sample Selection
Data Sharing Statement
References
- 1.Vervoort D, Lee G, Ghandour H, et al. Global cardiac surgical volume and gaps: trends, targets, and way forward. Ann Thorac Surg Short Rep. 2023;2(2):320-324. doi: 10.1016/j.atssr.2023.11.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Hensley NB, Cho BC, Visagie M, Lester LC, Abernathy JH III, Frank SM. How do I audit intraoperative blood component utilization in cardiac surgery? Transfusion. 2019;59(10):3058-3064. doi: 10.1111/trf.15399 [DOI] [PubMed] [Google Scholar]
- 3.Bartoszko J, Martinez-Perez S, Callum J, Karkouti K; FIBRES Study Investigators . Impact of cardiopulmonary bypass duration on efficacy of fibrinogen replacement with cryoprecipitate compared with fibrinogen concentrate: a post hoc analysis of the Fibrinogen Replenishment in Surgery (FIBRES) randomised controlled trial. Br J Anaesth. 2022;129(3):294-307. doi: 10.1016/j.bja.2022.05.012 [DOI] [PubMed] [Google Scholar]
- 4.Myles PS, Smith JA, Forbes A, et al. ; ATACAS Investigators of the ANZCA Clinical Trials Network . Tranexamic acid in patients undergoing coronary-artery surgery. N Engl J Med. 2017;376(2):136-148. doi: 10.1056/NEJMoa1606424 [DOI] [PubMed] [Google Scholar]
- 5.Pandey S, Belanger GA, Rajbhandary S, et al. A survey of US hospitals on platelet inventory management, transfusion practice, and platelet availability. Transfusion. 2021;61(9):2611-2620. doi: 10.1111/trf.16561 [DOI] [PubMed] [Google Scholar]
- 6.Rajbhandary S, Shmookler A, Cohn CS, et al. Hospital transfusion service operations during the SARS-CoV-2 pandemic: lessons learned from the AABB hospital survey in preparation for the next infectious disease outbreak. Transfusion. 2021;61(11):3129-3138. doi: 10.1111/trf.16643 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Henderson RA, Judd M, Strauss ER, et al. Hematologic evaluation of intraoperative autologous blood collection and allogeneic transfusion in cardiac surgery. Transfusion. 2021;61(3):788-798. doi: 10.1111/trf.16259 [DOI] [PubMed] [Google Scholar]
- 8.Barile L, Fominskiy E, Di Tomasso N, et al. Acute normovolemic hemodilution reduces allogeneic red blood cell transfusion in cardiac surgery: a systematic review and meta-analysis of randomized trials. Anesth Analg. 2017;124(3):743-752. doi: 10.1213/ANE.0000000000001609 [DOI] [PubMed] [Google Scholar]
- 9.Casselman FPA, Lance MD, Ahmed A, et al. ; EACTS/EACTAIC/EBCP Scientific Document Group . 2024 EACTS/EACTAIC Guidelines on patient blood management in adult cardiac surgery in collaboration with EBCP. Eur J Cardiothorac Surg. 2025;67(5):ezae352. doi: 10.1093/ejcts/ezae352 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Salenger R, Arora RC, Bracey A, et al. Cardiac surgical bleeding, transfusion, and quality metrics: joint consensus statement by the enhanced recovery after surgery cardiac society and society for the advancement of patient blood management. Ann Thorac Surg. 2025;119(2):280-295. doi: 10.1016/j.athoracsur.2024.06.039 [DOI] [PubMed] [Google Scholar]
- 11.Goldberg J, Paugh TA, Dickinson TA, et al. ; PERForm Registry and the Michigan Society of Thoracic and Cardiovascular Surgeons Quality Collaborative . Greater volume of acute normovolemic hemodilution may aid in reducing blood transfusions after cardiac surgery. Ann Thorac Surg. 2015;100(5):1581-1587. doi: 10.1016/j.athoracsur.2015.04.135 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Jacobs JW, Diaz M, Arevalo Salazar DE, et al. United States blood pricing: a cross-sectional analysis of charges and reimbursement at 200 US hospitals. Am J Hematol. 2023;98(7):E179-E182. doi: 10.1002/ajh.26940 [DOI] [PubMed] [Google Scholar]
- 13.Henderson RA, Mazzeffi MA, Strauss ER, et al. Impact of intraoperative high-volume autologous blood collection on allogeneic transfusion during and after cardiac surgery: a propensity score matched analysis. Transfusion. 2019;59(6):2023-2029. doi: 10.1111/trf.15253 [DOI] [PubMed] [Google Scholar]
- 14.Mladinov D, Padilla LA, Leahy B, et al. Hemodilution in high-risk cardiac surgery: laboratory values, physiological parameters, and outcomes. Transfusion. 2022;62(4):826-837. doi: 10.1111/trf.16844 [DOI] [PubMed] [Google Scholar]
- 15.Geube M, Sale S, Bakdash S, et al. Prepump autologous blood collection is associated with reduced intraoperative transfusions in aortic surgery with circulatory arrest: a propensity score-matched analysis. J Thorac Cardiovasc Surg. 2022;164(5):1572-1580.e5. doi: 10.1016/j.jtcvs.2021.01.029 [DOI] [PubMed] [Google Scholar]
- 16.Muñoz M, Acheson AG, Auerbach M, et al. International consensus statement on the peri-operative management of anaemia and iron deficiency. Anaesthesia. 2017;72(2):233-247. doi: 10.1111/anae.13773 [DOI] [PubMed] [Google Scholar]
- 17.Shander A, Hofmann A, Ozawa S, Theusinger OM, Gombotz H, Spahn DR. Activity-based costs of blood transfusions in surgical patients at four hospitals. Transfusion. 2010;50(4):753-765. doi: 10.1111/j.1537-2995.2009.02518.x [DOI] [PubMed] [Google Scholar]
- 18.Shander A, Ozawa S, Hofmann A. Activity-based costs of plasma transfusions in medical and surgical inpatients at a US hospital. Vox Sang. 2016;111(1):55-61. doi: 10.1111/vox.12386 [DOI] [PubMed] [Google Scholar]
- 19.Hofmann A, Ozawa S, Shander A. Activity-based cost of platelet transfusions in medical and surgical inpatients at a US hospital. Vox Sang. 2021;116(9):998-1004. doi: 10.1111/vox.13095 [DOI] [PubMed] [Google Scholar]
- 20.Austin PC. A comparison of 12 algorithms for matching on the propensity score. Stat Med. 2014;33(6):1057-1069. doi: 10.1002/sim.6004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Monaco F, Lei C, Bonizzoni MA, et al. ; ANH Study Group . A randomized trial of acute normovolemic hemodilution in cardiac surgery. N Engl J Med. 2025;393(5):450-460. doi: 10.1056/NEJMoa2504948 [DOI] [PubMed] [Google Scholar]
- 22.Hensley NB, Gyi R, Zorrilla-Vaca A, et al. Retrograde autologous priming in cardiac surgery: results from a systematic review and meta-analysis. Anesth Analg. 2021;132(1):100-107. doi: 10.1213/ANE.0000000000005151 [DOI] [PubMed] [Google Scholar]
- 23.Stubbs JR, Homer MJ, Silverman T, Cap AP. The current state of the platelet supply in the US and proposed options to decrease the risk of critical shortages. Transfusion. 2021;61(1):303-312. doi: 10.1111/trf.16140 [DOI] [PubMed] [Google Scholar]
- 24.Slater M. Olympics cycling: marginal gains underpin Team GB dominance BBC Sport. August 8, 2012. Accessed April 30, 2025. https://www.bbc.com/sport/olympics/19174302
- 25.Cahill CM, Alhasson B, Blumberg N, et al. Preoperative anemia management program reduces blood transfusion in elective cardiac surgical patients, improving outcomes and decreasing hospital length of stay. Transfusion. 2021;61(9):2629-2636. doi: 10.1111/trf.16564 [DOI] [PubMed] [Google Scholar]
- 26.Althoff FC, Neb H, Herrmann E, et al. Multimodal patient blood management program based on a three-pillar strategy: a systematic review and meta-analysis. Ann Surg. 2019;269(5):794-804. doi: 10.1097/SLA.0000000000003095 [DOI] [PubMed] [Google Scholar]
- 27.Nishioka N, Ichihara N, Bando K, et al. Body mass index as a tool for optimizing surgical care in coronary artery bypass grafting through understanding risks of specific complications. J Thorac Cardiovasc Surg. 2020;160(2):409-420.e14. doi: 10.1016/j.jtcvs.2019.07.048 [DOI] [PubMed] [Google Scholar]
- 28.Henderson R, Chow JH, Tanaka KA. A bridge to bloodless surgery: use of hemoglobin-based oxygen carrier for anemia treatment and autologous blood preservation during redo pulmonic valve replacement. J Cardiothorac Vasc Anesth. 2019;33(7):1973-1976. doi: 10.1053/j.jvca.2018.09.036 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Propensity Score Model Parameter Estimates
eTable 2. Post-Match Variables in the Propensity Score Model
eTable 3. Parameter Estimates for Model of ANH (Yes/No) for Any Blood Component Transfusion
eTable 4. Parameter Estimates for Model of ANH volume categories (3-level) for Any Blood Component Transfusion
eFigure 1. Study Population and Sample Selection
Data Sharing Statement
