Central Line Access Is Predictive of Diagnostic Blood Loss and Transfusion in the Surgical Intensive Care Unit.

BACKGROUND
Most patients in the surgical intensive care unit (SICU) have anemia and undergo extensive diagnostic laboratory testing (DLT). Consequently, patients undergo RBC transfusion, and many are discharged with anemia, both of which are associated with poorer outcomes.


OBJECTIVE
To characterize DLT blood loss in the SICU.


MATERIALS AND METHODS
We performed a 1-year retrospective study of 291 patients admitted to a SICU. The number of draws, average volume, and estimated discard volume were recorded, along with clinical and laboratory findings.


RESULTS
Patients who underwent greater amounts of DLT had lower hemoglobin levels at discharge (P ≤ .001). Admissions requiring central venous catheter (CVC) access (49.8%) demonstrated significantly higher DLT draws and rates of transfusion.


CONCLUSION
Findings from this study suggest that DLT blood loss contributes to anemia in the SICU, and that the presence and duration of CVC leads to increased testing, anemia, and RBC transfusion.


Introduction
Diagnostic laboratory testing (DLT) plays an integral role in the clinical management of critically ill patients. While DLT provides invaluable information, helping guide diagnostic, prognostic and therapeutic decisions volume lost due to high testing burden has been shown to contribute to the development of anemia and may lead to an increased rate of transfusion. [1][2][3][4] An estimated 16.2 laboratory tests are collected daily from ICU patients and account for a daily DLT-associated blood loss of 15 to 100 mL. 2,4−9 In the ICU, blood needed for DLT is frequently collected from a central venous access catheter (CVC).
Approximately half of patients in the ICU will undergo placement of a CVC. 10 The ease at which CVC's permit collection of DLT and the clinical acuity of the patients that often require CVC appears to promote increased access, with a an approximately 30% increase in testing and a 40% increase blood volumes collected in patients with CVCs relative to those without CVCs. 11,12 Ultimately, an increase in DLT and increased DLT-associated blood loss has been shown to correspond with a higher incidence of anemia. [1][2][3][4] Anemia is a common development in the critical care setting, with a reported incidence of anemia in 98% of patients admitted to the ICU. 9 Suppressed erythropoiesis observed in critically ill patients occurs due to in ammation prompting iron sequestration and decreased erythropoietin activity. [13][14][15][16][17][18] Other etiologies include nutritional de ciencies, major surgical interventions, bleeding, LOS, and receipt of care at a teaching institution. 17,19,20 21-23 Together these factors lead to the development of anemia at discharge in ICU patients, with severe discharge anemia associating with a number of adverse outcomes, including increased 30-day mortality rates and readmissions, representing a signi cant source of healthcare costs. 13,21 The prevalence of anemia in the critically ill often prompts RBC transfusions, occurring in 26.3-50% of ICU patients. [24][25][26] The high rate of RBC transfusions is concerning given the observed relationship between liberal RBC transfusions and poor outcomes, such as a higher incidence of serious health-careassociated infections and increased LOS. 25,27,28 Similarly, mortality rates are appreciably higher in transfused ICU patients relative to non-transfused patients, 30% vs. 19.6%, respectively. 26 Measures aimed at mitigating factors that contribute to both the incidence of anemia and the extent of RBC transfusions have the potential to signi cantly improve patient outcomes in the critically ill. Our study aimed to characterize DLT-associated blood loss, the impact of CVC placement, and their association with anemia and rate of RBC transfusion.

Design and Setting:
This retrospective study was conducted at the University of Texas Medical Branch (UTMB) in Galveston, Texas, between January 1, 2017, and December 31, 2017, and was approved by the local institutional review board. Data from hospitalization was recorded and no follow-up data was obtained. Cost analysis was performed at University of Texas Southwestern at Parkland Hospital in October of 2021 and did not involve the use of patient data. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines were utilized in preparation of this manuscript. 29 Study Population: Our study population comprised consecutive adult patients admitted to the SICU. Patients <18 years of age as well as vulnerable populations were excluded from the study. The study size was a convenience sample derived from the time period reviewed.

Data Collection:
Electronic medical records were manually reviewed with the following data obtained: patient demographics and clinical information (Table 1), laboratory results, progress notes, and discharge summaries. Data collection was performed by physicians, recorded in standardized de-identi ed abstraction sheets, and met the requirements of the Health Insurance Portability and Accountability Act. To account for the acuity of illness we calculated the Charlson Comorbidity Index (CCI) for each patient as previously described. 30  To estimate the total blood volume collected for a given laboratory tests, tube volumes were con rmed by injecting with water and then measuring nal volume into tubes. The number and source (i.e., peripheral access, CVC) of DLT was identi ed in the patient's electronic medical records. The product of the average 7-day blood volume and the number of each test type was then calculated to estimate the daily and total blood volumes collected for a given patient. Blood discarded from CVC, arterial lines, and peripherally inserted central catheter (PICC) were estimated based on internal nursing protocol requirements for minimum discard volume.
The following information was also recorded: incidence of anemia ( (p=<0.001). The cumulative admission blood loss was predictive of the magnitude of anemia at discharge, with the largest diagnostic-associated blood loss group corresponding to a discharge Hb of 8.6 g/dL versus the lowest blood loss group being associated with a Hb of 9.7 g/dl (p=<0.001) (Figure 1).

Cvc Vs Non-cvc
Patients with and without CVC had similar length of stay and CCI (Table 2). However, patients with CVC access had signi cantly higher number of total blood draws, total blood drawn or discarded for access and DLT, and rate of transfusion. Patients with CVC require clearing of access with a discard volume of 204 mL/hospitalization. Rates of anemia on discharge were signi cantly higher in the CVC group. Also, nadir Hb levels and discharge Hb were signi cantly lower in patients with CVC placement. The number of total blood draws per day was signi cantly higher in patients with a longer duration of CVC (Figure 2), and total number of draws was signi cantly higher in patients with CVC versus those without. The volume of combined DLT-associated blood loss was predictive of receipt of RBC transfusion, independent of a surgical procedure or acuity of illness as complicated by co-morbidities by CCI (Figure 3), though the range of CCI was greater in patients with prolonged CVC access ( Figure 4).  Patients with and without anemia on admission had similar length of stay and CCI (Table 3). Admission anemia signi cantly associated with lower Hb nadir, Hb at discharge, and rate of anemia at discharge. Pre-existing anemia was not predictive of transfusion.  (Table 4). Surgical patients had higher rates of CVC placement, lower Hb nadir, and lower Hb at discharge. Receipt of surgery was not predictive of anemia at discharge or non-operative transfusion.

Discussion
The ndings of our study suggest that the total volume of DLT-associated blood loss contributes to both the incidence and magnitude of anemia and was associated with more frequent RBC transfusions in critically ill patients. Likewise, the severity of anemia correlated with the volume of blood drawn ( gure 1).
Furthermore, CVC access was predictive of increased total blood loss, transfusion, and anemia. The observations of our investigation correspond with those reported in previous studies. 1,4−9 CVC placement associated with an increase in blood draws for laboratory testing and a resultant increase in iatrogenic blood loss, both for DLT as well as line clearance blood discards. While the critically ill require extensive DLT to guide therapy, the number of accesses as well as the utilization of blood conserving devices (BCD) to reduce discards represent modi able factors. BCD use may be especially impactful as there was a signi cant amount of cumulative blood loss associated with line clearance, 204 mL/admission, which nears the red cell volume of 1 unit of RBCs. A recent meta-analysis by Whitehead et al., demonstrated that utilization of BCDs were associated with a reduction in iatrogenic blood loss as well as the use of pediatric tubes. 5 While the overall rate of transfusion was 32% (95/299), CVC access was predictive of a higher rate compared to those without CVC access. Transfusion increases morbidity and mortality amongst ICU patients and thus, DLT blood loss represents an area which could be impacted with implementation of common-sense blood saving interventions, such as BCD. We found that CVC line placement >7-days was predictive of a higher rate of RBC transfusions. ICU services often have goals to remove CVCs as quickly as possible, to reduce the risk of central line associated bloodstream infection (CLABSI), which are associated with high mortality, and our ndings suggest reduction in DLT blood loss represents a further motive for early CVC removal. 33 One critique of BCDs is an increased cost. A charge analysis performed demonstrated bedside CVC placement was roughly equivalent to the charge for a single RBC transfusion ($1,340 vs. $1,450). Though many services place lines at bedside there has been a shift in some centers with the majority being placed by interventional radiology, which we show to be signi cantly more expensive. Although determining clear pricing information is challenging, utilization of BCDs would prevent doubling the cost of having a bedside CVC if an RBC transfusion is not required.
CVC placement led to lower Hb nadirs and discharge Hb levels. This is impactful, as the body increases cardiac output and shunts blood away from visceral organs through compensatory mechanisms in response to anemia, which may lead to further end organ stress or failure. Also, severe discharge anemia has been shown to be predictive of anemia on long term follow-up for ICU patients, further contributing to readmission, need for transfusion, and poor general health as an outpatient. 13,16 Under physiologic conditions the body can readily produce su cient red cell mass for maintenance of Hb levels, ~10-20 mL per day, and though this number may increase exponentially in states of anemia with adequate iron stores, the critically ill patient has been shown to have reduced red cell production, making any blood loss likely more impactful. 18,34−37 Many ICU patients are anemic on presentation with nutritional de ciency being a contributing factor in 13% of ICU patients as follows: 2% B12 de cient, 2% folate de cient, and 9% iron de cient. 17 To that end, each ml of blood contains 1 mg of iron, and while patients generally have iron stores of 300-1000 mg, this iron may be rendered inaccessible due to an ongoing in ammatory response. 34,38,39 Moreover, though healthy blood donors have been shown to have a maximum absorption between 3-5 mg/day iron it is likely that critically ill patients have suboptimal absorption of nutrients. 40 Erythropoietin is produced predominantly in the kidney with a small percentage produced in the liver, and work shows ICU patients have both reduced production of EPO as well as a blunted response. 41,42 Ongoing blood draws represent an identi able loss of red cell mass and iron which the critically ill patient cannot effectively replace, further contributing to iatrogenic anemia.
While we observed a total mean daily loss of blood of 20.0 mL, a volume below that observed ~50-100 mL in other studies, this volume is actually equivalent to steady state RBC production in healthy individuals. 4-7, 36,43 A mathematical model by Lyon et al. suggested that loss of 53 mL per day in healthy patients with adequate iron stores would lead to anemia in approximately 40 to 70 days. 44 Conversely, critically ill patients with compromised RBC production would demonstrate anemia in approximately 9 to 14 days. 44 The observed SICU LOS in our study 15 days, and though 20 mL was signi cantly less than 53 ml, 95.3% of patients were anemic at time of unit discharge.
One limitation of the study that many patients had both CVC and PICC placement. As such those with only CVC were compared with PICC and those with both PICC and CVC to compare discard volumes. As a single predictor both DLT-associated and total blood loss was signi cantly lower in those without CVC, and CVC placement alone was predictive of increased blood draws.
As this is a retrospective analysis, we could not actively monitor these patients or obtain additional studies that may have been informative, such as nutritional or in ammatory markers. Moreover, these are complex, critically ill patients and numerous factors, as discussed above, contribute to the incidence of anemia, though we attempt to characterize this with CCI values and by comparing surgical and nonsurgical patients. Our group does show a contemporary 'baseline' of iatrogenic blood loss in an SICU setting, which may be generalizable to other SICU services. Likewise, we also showed lower rates of transfusion compared to prior work, showing some movement toward more conservative transfusion practice. The goal of a "bloodless" ICU admission without transfusion should still be pursued as RBC transfusion in this population is associated with higher rates of mortality, increased LOS, and ongoing anemia as an outpatient. 13,28 Several limitations including the accuracy of data recorded within the electronic medical records and selection bias. Moreover, the decision to transfuse RBCs was left to the discretion of the treating physician and therefore the rationale for RBC transfusions was not always clear.
Finally, chargemaster values were used as opposed to direct costs to the hospital or patient.
Chargemaster rates have been shown to have high variability and are likely not predictive of cost to patient, but may provide some insight into billing for these procedures. 45 Conclusions While onset or persistence of anemia in this context is multifactorial, an association is clear between DLT, CVCs and hospital acquired anemia. ICU patients especially require numerous laboratory tests given the high acuity of care; however, physicians should be aware that this increased testing associates with higher rates of anemia and the need for RBC transfusion. Given the associated increase in increase morbidity and mortality associated with hospital acquired anemia and transfusion, efforts should be made to reduce unnecessary DLT and associated adverse events to improve quality of care.

Declarations
Con ict of Interest: The authors declare that they have no known competing nancial interests or personal relationships that could have appeared to in uence the work reported in this paper.
Ethics approval and consent to participate IRB approval was obtained for main study. Chargemaster analysis was performed on publicly available data.

Consent for publication
This was a retrospective study and no individual consent was obtained.

Availability of data and material
Data is available and is in the possession of the rst and senior author. De-identi ed data may be provided by contacting the senior author.

Competing interests
There are no relevant competing interests.

Funding
No funding was obtained for this retrospective chart review study.   Charlson Comorbidity Index (CCI) comparison amongst patients with CVC placement.