Bleeding Complications in Ultrasound-Guided Central Venous Access in Patients with Severe Thrombocytopenia: A Propensity Score-Based Analysis

Purpose: To study the occurrence of bleeding complications in patients with severe thrombocytopenia (platelet count <20x10 3 /µL) subjected to ultrasound-guided central venous access (UG-CVA) while receiving or not routine prophylactic platelet transfusion (PPLT). Research Question: What is the frequency of bleeding complications related to the placement of ultrasound-guided central venous access in patients with severe thrombocytopenia between 2011 and 2019 at high complexity hospital? Methods: A total of 221 patients with severe thrombocytopenia subjected to UG-CVA from January 2.011 to November 2.019 were selected. They were classied as positive (P-PPLT) or negative (N-PPLT) recipients of PPLT. Then, P-PPLT (n=72) were 1:1 propensity matched to N-PPLT based on catheter diameter, anatomical insertion site, presence of hematologic malignancy, absolute platelet count and whether the health care provider performing the procedure was an attending or a trainee. Bleeding complications were graded according to the Common Terminology Criteria for Adverse Events (CTCAE) score and adapted to central venous catheter placement. A logistic regression analysis was then performed using “bleeding complications” as a binary compound outcome of major (Grades 3-4) and minor bleeding (Grades 1-2) vs. no bleeding. Results: Seventy-two patients were classied as P-PPLT, while 149 as N-PPLT. No grades 3-4 of bleeding events were identied in the entire population. No signicant differences were observed between N-PPLT and P-PPLT for bleeding Grades 1-2 in both pre-matched (53[35.5%] vs. 26[36.1%], p=0.90) and propensity-matched populations (27[37.5%] vs. 26[36.1%], p=0.80). Logistic regression demonstrated that PPLT did not inuence any bleeding complication (OR 0.9, 95%CI 0.42-1.92, p=0.791) Conclusions: Bleeding complications related to central venous catheterization in acutely ill patients with severe thrombocytopenia are not inuenced by routine prophylactic platelet transfusion when catheter placement is performed under ultrasound guidance.


Introduction
Central venous catheterization is a procedure frequently needed in acutely ill patients. In many cases, acute illness is related with thrombocytopenia and prolonged coagulation times, which increase the risk of bleeding during invasive procedures. In general, rates of bleeding complications associated with insertion of central venous lines range between 1% and 10% [1][2][3][4][5]. Nevertheless, the overall risk of major and minor bleeding associated with a central line placement in patients with coagulation abnormalities is highly variable, with bleeding rates ranging between 0 and 32% [6-8] and no deaths were directly attributed to periprocedural bleeding complications [9][10][11][12]. Although some authors claim there is no conclusive evidence about the optimal platelet count before central venous catheter insertion [13], observational studies suggest that counts > 50 x10 3 /µL could be safe. Current guidelines recommend prophylactic platelet transfusion (PPLT) before central venous placement in patients with platelet counts < 20 x10 3 /µL [6,14] or 50 x10 3 /µL [15][16]. Nevertheless, the strength of such recommendations is weak, and these have been widely based on retrospective observational studies including a very low number of patients with counts < 20 x10 3 /µL [7,[17][18][19][20][21][22]. Thus, using a predetermined platelet count threshold to routinely use PPLT before central venous catheter placement remains highly debatable and platelet transfusion is not free of risk [23][24].
Ultrasound guidance for central venous catheter placement signi cantly reduces bleeding complications [2,5,[25][26] and its use is recommended by international guidelines [27][28][29]. Nevertheless, there is not clear information about the safety of ultrasound-guided central venous access (UG-CVA) in severe thrombocytopenic patients without using routine prophylactic platelet transfusion. Thus, we aim to evaluate the occurrence of bleeding complications associated with UG-CVA in acutely ill patients with platelet counts < 20 x10 3 /µL, receiving or not a prophylactic platelet transfusion.

Study population / design
We retrospectively identi ed all adult patients ≥ 18 years old with severe thrombocytopenia (platelet count < 20 x10 3 /µL) of any cause who were subjected to UG-CVA from January 2.011 to November 2.019, in a fourth level university hospital (Fundación Valle del Lili, Cali, Colombia). The local ethical and research committee involving human beings approved this study (Protocol number 1279, IRB/EC approval letter number 210-2018, Fundación Valle del Lili, Cali, Colombia. In agreement with the nature of the study, informed consent was waived.
Clinical and laboratory records were carefully reviewed by the investigators. Patients with platelet counts > 20 x10 3 /µL and those in whom central venous cannulation was not ultrasound-guided, were not included for the analysis. Complete demographics, transfusion requirements, catheter size, insertion site, whether the person performing the invasive procedure was an attending or an in-training physician and bleeding complications were collected.
This study was adhered to the standards of the STROBE (Strengthening the Reporting of Observational studies in Epidemiology) guidelines for the report of observational studies [30] and a STROBE statement is provided in Supplementary digital content.

Outcomes
Primary outcome was the incidence of a combination of major and minor bleeding complications associated with UG-CVA in patients with platelet count < 20 x10 3 /µL receiving or not a prophylactic platelet transfusion.
Bleeding complications were graded according to the Common Terminology Criteria for Adverse Events (CTCAE) score and adapted by Zeidler for CVC placement [22,[31][32]. A major bleeding event was de ned as that requiring any surgical / endovascular / percutaneous intervention, red cell transfusion or associated hemodynamic instability (Grades 3-4). Minor bleeding was de ned as oozing from the skin insertion site or hematoma that did not required intervention and resolved with manual compression (Grades 1-2). Additionally, hemoglobin values were analyzed before and 24 hours after the procedure. A decrease in hemoglobin level ≥ 1.5g/dL judged to be associated with the catheter insertion, was also considered a major bleeding event.

Statistical analysis
Continuous variables were summarized as median and inter-quartile ranges according to their nonnormal distribution. All qualitative variables were expressed as frequencies and percentages. Nonnormally distributed continuous variables were compared by using the Wilcoxon signed-rank test.
Categorical variables were compared by using the Pearson χ 2 test, the Fisher exact test, or the McNemar test as appropriate. The decision to use a PPLT was always taken by the attending physician according to his clinical judgment. Since patients undergoing UG-CVA were not randomly assigned either to positive prophylactic platelet transfusion (P-PPLT) or negative (N-PPLT), a formal causal inference regarding such platelet transfusion and clinical outcomes of interest is not possible. Thus, we generated a propensity score for each patient using a logistic regression model in order to predict the probability (propensity) of receiving or not PPLT before UG-CVA. Such score was a composite of catheter diameter (13.5Fr vs. 5-8 Fr), venous access site (venous jugular, subclavian or femoral), presence of hematologic malignancy, absolute platelet count, use of anticoagulant and antiaggregant therapy, physician performing the catheter insertion (emergency medicine vs. intensive care specialist and senior vs. in-training physician).
After tting the propensity score, a nearest neighbor-matching algorithm extracted 1:1 matched pairs of P-PPLT and N-PPLT subjects.
A logistic regression analysis was then performed using "bleeding complications" as a binary compound outcome of major (Grades 3-4) and minor bleeding events (Grades 1-2) according with the CTCAE classi cation for bleeding adapted for central venous catheter placement. In addition, a clearly procedurerelated decrease in hemoglobin levels > 1.5g/dL or an unexplained drop in hemoglobin > 1.5g/dL within the next 24-hours were also considered a major bleeding complication. Logistic regression was conducted including as covariables: site of insertion, type of catheter (gauge), hemato-oncologic disease, pH, previous use of anti-aggregants or anticoagulants, procedural performer training level (in-training vs. faculty); and, platelet count: < 5 x10 3 /µL, between 5 and 10 x10 3 /µL, or > 10 x10 3

Results
Between January 2011 and November 2019, a total of 16658 patients underwent central venous catheterization. Of these, 16170 patients had a platelet count ≥ 20 x10 3 /µL and 167 were not inserted under ultrasound-guidance. Finally, 221 patients were included for the analysis. Selection of patients is presented in the Fig. 1. A description of demographics and general characteristics of the overall population is provided in the Table 1. Major bleeding complications (CTCAE grades 3-4) were not observed in the entire cohort, while minor bleeding complications (CTCAE Grades 1-2) were detected in 79 (35.7%) patients. Although 6 (2,7%) patients showed a drop ≥ 1,5g/dL in hemoglobin levels into the 24 hours of catheter placement, a careful review of medical charts revealed that such a decrease was apparently not associated to the catheter insertion itself but mostly due to the own disease course or other bleeding complications. After the propensity-matched procedure, 72 patients were classi ed as P-PPLT and other 72 as N-PPLT (Table 2). General characteristics of this matched-cohort are showed in the   Remarkably, no major bleeding complications were detected in the entire cohort and the total number of minor bleeding events was quite similar to the previously reported in the literature, which highlights the role of ultrasound as a tool to increase safety in these clinical conditions. These ndings were consistent even in the subgroup of patients with platelet counts below 10 x10 3 /µL. In addition, subclavian access was not associated with an increased rate of bleeding complications compared to the jugular site. Interestingly, procedures without prophylactic platelet transfusion were most commonly performed by emergency medicine physicians, which probably re ects a higher frequency of emergent central venous cannulations in this setting.
This study has several limitations. First, its retrospective nature and, therefore, the lack of control by randomization and blinding might limit the external validity of conclusions. Admittedly, although propensity scores were constructed incorporating baseline characteristics likely in uencing the decision for prophylactic platelet transfusions (other than thrombocytopenia itself), other non-identi able factors might not have been included. In addition, the relatively small sample size of this cohort might introduce a risk of missing important differences at the baseline which might contribute to the absence of difference between the P-PPLT and N-PPLT groups. Second, we were not able to identify if the decision on "non-transfusion" relied on some particular attending physicians, which could constitute a potential factor of confusion. Nevertheless, we speculate that the lower occurrence of prophylactic transfusions ordered by emergency physicians respond to a higher frequency of emergency situations not captured by the retrospective nature of our observation. Third, although ultrasound-guided catheterization is an apparently safe procedure in severe thrombocytopenic patients, appropriate training and developed skills are fundamental to reproduce good results. In our hospital, there is an established ultrasound training program for emergency department and critical care staff, so these ndings may not be replicable in other environments.
Fourth, even though our results suggest the apparent safety of non-routine prophylactic transfusion in severe thrombocytopenic patients subjected to venous catheterization under ultrasound guidance, future prospective research efforts are necessary to con rm such results.

Conclusion
Bleeding complications related to central venous catheterization in acutely ill patients with severe thrombocytopenia are not in uenced by routine prophylactic platelet transfusion when catheter placement is performed under ultrasound guidance.
Our ndings reinforce the idea about the safety of central venous catheter placement when appropriate ultrasound guidance is used in patients with severe thrombocytopenia and provides more data on the effect of prophylactic platelet transfusion for bleeding complications in this subset of patients.

Declarations
Funding: We have not received funding for this study.
Con icts of interest: The authors have no con icts of interest to declare.
Ethics approval and consent to participate: The local ethical and research committee involving human beings approved this study (Protocol number 1279, IRB/EC approval letter number 210-2018), Fundación Valle del Lili, Cali, Colombia. In agreement with the nature of the study, informed consent was waived.
Consent for publication: Not applicable.
Availability of data and materials: The data that support the ndings of this study are available from the systematic database (BD clinic) of our hospital Fundación Valle del Lili but, restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission from Fundación Valle del Lili.
Acquisition, analysis or interpretation of data: All authors.
Critical revision of the manuscript for important intellectual content and approved the version to be published: All authors.