Use of Low-Dose Platelets in Actively Bleeding Patients : A Retrospective Analysis of a Cardiac Surgery Cohort

This study was conducted at the University of Texas Medical Branch in Galveston, a level 1 trauma center that supports 20 operating rooms (ORs) as well as multiple transplant programs. Use of low-dose platelets has been implemented by our institution’s blood bank in times of critical shortage since December 2021. Owing to the hospital’s location and distance from blood suppliers, significant delays in platelet shipments can occur. When faced with a critical shortage without immediate relief, low-dose platelet units are issued to patients meeting criteria for platelet transfusion in a manner that is semiblind to the clinical indication, meaning low-dose platelets are not specifically issued to a subtype of patients but rather to all patients meeting criteria for platelet transfusion.

This study was reviewed by our institutional internal review board and received an acknowledgment of nonregulatory activity.

We produce low-dose platelets when we reach a critical threshold. By internal protocol, low-dose units are preferentially produced from platelet units with yields of 5 × 10¹¹ platelets or greater per unit, although at our institution, units with these yields are rarely available. When yields are not available from the blood supplier, a minimum yield of 3 × 10¹¹ platelets per unit is assumed, thereby producing 2 units with a minimum yield of 1.5 × 10¹¹ platelets. Within our blood bank, a log of the estimated yield of all low-dose platelets produced is maintained. Review of this log for a 2-month period indicates that exact yields were known for 66 of the 176 units split (38%) of the low-dose units produced and assumed for 110 of the 176 units split (62%). The mean estimated yield of low-dose units with a known yield was 1.89 × 10¹¹ platelets per unit, with a range of estimated platelet yields from 1.5 × 10¹¹ to less than 3.0 × 10¹¹. A histogram plot of the known yields of low-dose units revealed that 62 of the 66 platelet units (94%) in this analysis had yields of 2.2 × 10¹¹ or less (histogram not shown).

To produce a low-dose unit, the platelet unit is weighed and then joined in a sterile fashion with a sterile joiner (Terumo TSCD-II, Terumo BCT, Lakewood, Colorado) to an accessory platelet storage set (item 70030, Terumo BCT). The accessory platelet storage set is gas permeable, allowing the split unit to keep the original expiration date. Platelets are allowed to pass from the original unit to the accessory bag until ∼50% weight of the original unit is obtained. The bags are then clamped, and the connecting tube is sealed and divided by using a tube sealer (Composeal, Fresenius Kabi, Oberursel, Germany). Both units are then given updated product codes beginning with either “A0” or “B0” to allow appropriate tracking in our electronic medical record, and both units are also given a colored tag reading “Low Volume.”

During the time of the study, our institution prospectively triaged blood product units. However, no platelet orders from ORs were triaged until a threshold number of 5 platelet units per case was reached, which did not occur for any surgical cases. Similarly, within the ORs, requests for RBCs, FFP, and Cryo are not triaged.

To identify our study population, we obtained all instances in which a platelet was dispensed to the ORs between January 1, 2022, and June 30, 2023. This population was sorted by specialty, and all cardiac cases were identified. A total of 75 cardiac cases that used platelets were identified, all of which occurred on CPB. For reference, our institution performed 417 index cardiothoracic cases during the study period. At our institution, patients are instructed to hold aspirin therapy the morning of surgery, and to discontinue clopidogrel therapy 5 days before surgery. Of the 417 cases performed by our cardiothoracic surgeons, 17 (4%) had clopidogrel within 5 days of surgery.

Patients who underwent a cardiac procedure were then subdivided into 2 main cohorts: those who received only whole-dose platelet units while in the OR, and those who received only low-dose platelet units while in the OR. Patients with mixed-dose platelet units were not included in the analysis owing to the inherent bias of requiring 2 or more platelet transfusions to be included in that cohort. To gauge surgical complexity and ensure balanced bleeding risk in the patient population, the procedure type/indication for surgery was reviewed by a cardiac surgeon. In cases where multiple procedures were performed within a single case, the surgeon assessed which was considered the primary indication. Cases in both cohorts were sorted into 8 classes of procedures by risk for mortality and morbidity, in descending order: aortic dissection (Dissection), thoracic aortic aneurysm repair/aortic root aneurysm repair, pericardiectomy, left ventricular assist device placement, valve replacement/repair with coronary artery bypass graft (CABG), valve replacement/repair alone, CABG alone, and finally, reentry for bleeding, which had variable risk depending on the index surgery.

In some cases, a single patient received multiple cardiac surgeries during their hospital course. Each procedure was placed in the appropriate cohort depending on whether whole- or low-dose platelets were used. In total, we identified 37 separate operations on 37 individual patients in the whole-dose cohort, and 38 operations on 36 individual patients in the low-dose cohort.

Demographic information accurate for the date of surgery including age, weight, height, and biological sex was retrieved from the electronic medical record (EMR). Body surface area (BSA) was calculated by using the Du Bois formula.

Clinical complexity was assessed with an APACHE-II score, which estimates the risk of mortality upon admission to the intensive care unit (ICU). The calculation for the APACHE-II score involves age, sex, comorbidities, vital signs, and certain laboratory values, with a score more than 30 indicating the highest risk for mortality. The data for the APACHE-II score were retrieved from our EMR within 24 hours of postsurgical cardiac ICU admission.

We determined the total number of platelet transfusions intraoperatively, in which each transfusion was counted as 1.0, as well as the dose in units for other blood products. We note that for this portion of the analysis, each individual operation was included as a separate data collection point.

To encompass the immediate postoperative period, we obtained the total number of blood product units given during a 24-hour period from the start of surgery. This global value includes units given both intraoperatively and in recovery. At our institution, the time a patient leaves the OR is only intermittently logged, so an exact determination of the postoperative period was not feasible. In one instance in the low-dose cohort, a patient received 2 separate procedures back-to-back. In this case, the 24-hour units were counted only once. We note that in all other instances of a patient receiving multiple procedures, they occurred on different days, and their 24-hour product values were counted separately. The N for this portion of the analysis was 37 in the whole-dose cohort and 37 in the low-dose cohort.

EBL was taken from the clinical anesthesia notes for each case, although specific values were not always recorded. We obtained EBL for 35 procedures in the whole-dose cohort, and 36 procedures in the low-dose cohort.

Index cases, or the first surgery of an admission, were isolated from both cohorts with an N of 34 in the whole-dose platelet cohort and N of 35 in the low-dose platelet cohort. Each chart was manually reviewed for documentation of bleeding complications, including repeated surgeries for bleeding, thoracentesis for hemothorax, noted excessive bleeding in surgery, or noted bleeding complications in the postoperative period.

The main patient outcome was all-cause mortality within 30 days of discharge. For this analysis, we removed any patients who received 2 or more procedures that appeared in different cohorts during their admission. The total number of patients analyzed in the whole-dose cohort was 36 and in the low-dose cohort was 33.

All statistical analysis was performed in either Microsoft Excel or Microsoft Visual Studio Code using Python with SciPy and NumPy imported. A 2-tailed Student t test was used for continuous variables. Fisher exact test was used for analysis of biological sex distribution, all-cause mortality, and bleeding complication in index cases. A χ² distribution test was used for analysis of case type. Other basic statistical analysis included the arithmetic mean and error of the mean.

Two patient cohorts were generated, one comprising patients who solely received whole-dose platelets and the other with patients who solely received low-dose platelets during cardiac surgery (Table). The 2 cohorts were compared for the distribution of age (P = .94), biological sex (P > .99), mean weight (P = .85), and mean BSA (P = .96). There was no significant difference in any of these metrics, indicating that the demographics of the 2 cohorts were comparable. The clinical complexity of the 2 cohorts was compared with an APACHE-II score, which estimates the risk of ICU mortality. Again, there was no significant difference in mean APACHE-II score (P = .96). The complexity of the surgical procedures was then compared across cohorts, since this factor could influence the bleeding risk and/or number of transfusions during surgery. Eight classes of procedures were included in this analysis (Figure 1), and no significant difference was found (P = .58).

We then examined changes in transfusion requirements due to the use of low-dose platelets. Platelet transfusions, platelet dose, and the dose of other blood products during the surgery were analyzed during 2 periods, intraoperative and 24-hour perioperative. The intraoperative analysis revealed that mean platelet transfusions were not increased in the low-dose cohort (1.53 ± 0.11) versus the whole-dose cohort (1.62 ± 0.12) (Figure 2, A), indicating that anesthesiologists and surgeons used similar numbers of platelet transfusions during each case. Analysis of other blood products, used as a surrogate for bleeding,²²  revealed no significant increase in mean transfused dose of FFP (P = .08), Cryo (P = .81), or RBCs (P = .72) (Figure 2, B).

Surgeries can be associated with a delayed bleeding risk from incomplete hemostasis achieved during the operation.^23,24  Thus, a second time point, the 24-hour perioperative period, was analyzed. This time point would conceivably capture whether low-dose platelets resulted in increased postoperative bleeding. The 24-hour perioperative period was defined as the start time of surgery through 24 hours, such that the analysis would be inclusive of intraoperative transfusions. Analysis of the 24-hour perioperative time point similarly revealed no significant increase in platelet transfusions (low-dose, 1.92 ± 0.18; whole-dose, 1.73 ± 0.14; P = .41), or the dose of Cryo (P = .55) or RBC (P = .24), although there was a statistically significant increase in FFP dose of 0.48 units (P = .02), which may not be clinically significant (Figure 3, A and B).

While transfusion requirements and blood product usage are useful surrogates for bleeding, we also evaluated the quantity of bleeding from each surgical case. EBL recorded by anesthesiologists at the end of each surgical procedure was obtained from the EMR. We note that there is a small number of cardiac anesthesiologists at our institution who would have seen all patients in both cohorts, thus helping to control for estimation error in this metric. There was no significant increase in mean EBL in the low-dose cohort (953 ± 116 mL) versus the whole-dose cohort (904 ± 101 mL) (Figure 4, A; P = .76). While transfused products and EBL are important metrics for acute bleeding, we also sought to analyze any additional complications from the use of low-dose platelets by analyzing index cases for bleeding complications. We found that 5 of 34 index cases (15%) in the whole-dose platelet cohort had bleeding complications and 7 of 35 index cases (20%) in the low-dose cohort had bleeding complications, which is not statistically different in this data set (Figure 4, B; P = .75). Finally, we analyzed all-cause mortality within 30 days of discharge. We observed no significant increase in all-cause mortality, with 3 deaths among 33 patients in the low-dose platelet cohort, and 2 deaths among 36 patients in the whole-dose platelet cohort (Figure 4, C; P = .67). Causes of death for all patients in both cohorts can be found in the Supplemental Table (see the supplemental digital content at https://meridian.allenpress.com/aplm in the May 2025 table of contents).

In this study, we compared 2 cohorts of patients undergoing cardiac surgery during an 18-month period: those who received only whole-dose platelets in their procedure, and those who received only low-dose platelets during their procedure owing to shortages. In our analysis, we found no significant increase in mean platelet transfusions in either the intraoperative or 24-hour perioperative period (Figure 2, A, and Figure 3, A), despite the lower platelet doses used. We found no evidence of a significant increase in the requirement for other blood products (RBC, FFP, Cryo) in the intraoperative period, and only a small increase in the requirement for FFP of ∼0.5 units in the 24-hour perioperative period (Figure 2, B; and Figure 3, B). We note that in reducing the volume of platelets given, the volume of plasma transfused would also be reduced. Thus, the requirement for additional FFP in the low-dose platelet group may simply reflect that they are receiving less plasma from their platelet transfusions. Similarly, our analysis found no significant increase in other metrics of patient outcome, including EBL (Figure 4, A), bleeding complications from index cases (Figure 4, B), and all-cause mortality within 30 days of discharge (Figure 4, C). Together, these findings tentatively suggest that low-dose platelets are equivalent to whole-dose platelets when used during cardiac surgery.

Several limitations of this study should be noted. First, the number of patients in this study is small, with only 73 patients enrolled and only 75 procedures analyzed. Thus, this descriptive study can only suggest equivalent hemostasis, while larger prospective studies are needed to confirm these findings and definitively evaluate safety and efficacy. We also note the limitations of EBL for measurement of intraoperative blood loss, which suffers from lack of standardization. Additionally, while we found no statistically significant adverse outcomes in our analysis, we note the potential for trends toward increased blood product use in the perioperative period, increased all-cause mortality, and increased bleeding complication from index surgeries, which can only be definitively investigated in larger, prospective clinical trials. Second, this study was limited to patients undergoing cardiac surgery, all of whom had a specific type of bleeding. Further analysis is needed to determine whether the results of this study are applicable to other types of bleeding, such as gastrointestinal bleeding, obstetric complications, and coagulopathies, before these results can be truly generalizable. Third, this study was a retrospective review of outcomes that was unable to control for all variables. We note that the number of cardiac surgeons and anesthesiologists is low at our institution, which helps reduce differences in clinical practices and surgical experience, but other confounding factors could be present. A prospective study would help better control variables, although any ethical concerns around the potential for patient harm from using a low-dose product would need to be addressed. Fourth, this study represents the experience of a single institution, and expansion of the analysis to cardiac surgery cases at multiple institutions would help support the current findings.

Two strengths of this study, in particular, would need to be carried over into future large prospective trials. First, this study is unique in addressing outcomes in actively bleeding patients. While the efficacy of low-dose platelets in prophylaxis of bleeding is already established, their use in active bleeding or therapeutic applications still needs to be addressed. Second, in this study, low-dose platelet units were dispensed regardless of the clinical state of the patient. The pretransfusion platelet level and other clinical indicators were not assessed before the release of a low-dose unit, making this study relatively blinded. Carrying this aspect over into future analysis could be beneficial.

While the findings may point toward equivalent hemostasis for low-dose platelets, an alternative explanation is that fewer platelet transfusions are needed during cardiac surgery, as a reduction in platelet dose received did not appear to substantially change transfusion requirements and clinical outcomes in this study. We note that the mean platelet transfusions in our study were similar to the median of 2 reported in the literature,²¹  suggesting our hospital’s practice did not deviate from normal surgical standards. Cardiac surgeries often require use of CPB, which has previously been shown to reduce platelet function.^25–27  However, the clinical utility of platelets in cardiac surgery remains debatable,^28,29  with transfusion practices varying widely in different settings,³⁰  although there is agreement that prophylactic transfusion of platelets before CABG is not recommended in non-thrombocytopenic patients.³¹  At our institution, intraoperative transfusion decisions are made through a combination of clinical status, particularly active bleeding, and intraoperative readings of viscoelastic hemostatic assays. Given the common use of platelets during cardiac surgeries traditionally, a potential reassessment of the frequency and/or dose of platelets in cardiac surgery could be useful in reducing the demands placed on blood bank services.

A central question arises from this study: what dose of platelets is truly necessary for achieving hemostasis in a bleeding patient? Currently, platelets are not administered as a measured dose specific to a patient and clinical situation, but as a unit, which moreover does not have a uniform yield. While clinical trials like PLADO have evaluated specific BSA-based dosing for platelet prophylaxis,¹⁷  almost all platelet units given to adult patients are administered as an entire bag owing to limitations on infrastructure. With current technology, delivery of tailored platelet aliquot/doses would be slow and cumbersome for blood bank staff. Thus, the exact platelet dose at which a bleeding patient would receive benefit from a platelet transfusion remains unknown. Indeed, this study adds to a body of literature that suggests lower doses may be acceptable during operative cases, which echoes experiential findings in other countries with lower platelet yield targets and similar outcomes. Possible explanations for efficacy of platelets at lower doses include that most platelets quickly traffic to the bleeding site for hemostasis, and/or that only a minimum number of platelets are needed to overcome inactivation on an extracorporeal circuit, such that current doses provide a surplus of platelets.

This study also has implications for the US blood supply. Currently, blood suppliers may issue a low-dose platelet unit with reduced yield (2.5–2.9 × 10¹¹) to certain blood banks, which will often use these units for prophylactic purposes. With expansion and further study of the above findings, these low-yield platelet units could be regularly used for an expanded range of clinical situations, including in bleeding patients. Furthermore, these results could give confidence to blood banks that already practice splitting platelets that their patients may have similar clinical outcomes. Finally, issuing low-dose platelets could help many hospitals better handle shortages and delayed shipments of this precious resource.

In conclusion, low-dose platelets may be preliminarily equivalent to whole-dose platelets for use in cardiac surgeries, supporting the current practices of many institutions. We advocate for additional multicenter studies to replicate this work in cardiac surgeries and other bleeding patient populations to verify adequate hemostasis in diverse clinical situations. The practice of administering low-yield platelets or splitting traditional platelet units could greatly expand the availability of this important product in times of shortage and chronic inventory challenges.