Effect of Intra- and Postoperative Fluid and Blood Volume on Postoperative Pulmonary Edema in Patients With Intraoperative Massive Bleeding: Multi-Center Cohort Study Using Time Varying Analysis

Background In patients with massive bleeding during surgery, the effect of intra- and postoperative uid and blood volume on postoperative pulmonary edema is uncertain. The aim of this study is evaluating the occurrence risk relationship through time-varying analysis between postoperative pulmonary edema and intra- and postoperatively administered volume of uid and blood in patients with intraoperative massive bleeding. Methods This study retrospective and Patients with intraoperative massive bleeding ( ≥ of underwent a non-cardiac the The primary outcome was postoperative pulmonary edema Secondary outcomes were postoperative pulmonary edema with hypoxemia.


Abstract Background
In patients with massive bleeding during surgery, the effect of intra-and postoperative uid and blood volume on postoperative pulmonary edema is uncertain. The aim of this study is evaluating the occurrence risk relationship through time-varying analysis between postoperative pulmonary edema and intra-and postoperatively administered volume of uid and blood in patients with intraoperative massive bleeding.

Methods
This study is a retrospective cohort study and data was obtained from the clinical data warehouse at Hallym University Medical Center, a multi-institutional data registry of 5 hospitals of Hallym University.
Patients with intraoperative massive bleeding (≥40% of average blood volume) and who underwent a non-cardiac surgery at 5 hospitals between January 1, 2010, and December 31, 2019 were enrolled the study. The primary outcome was postoperative pulmonary edema occurrence within 72 hours after surgery. Secondary outcomes were postoperative pulmonary edema with hypoxemia.

Results
In total, 2090 patients were included in the postoperative pulmonary edema analysis, and 300 patients developed pulmonary edema within 72 hours after surgery. The postoperative pulmonary edema hypoxemia analysis with hypoxia included 1660 patients; pulmonary edema with hypoxemia occurred in 161 patients. The increase in the amount of red blood cells/average blood volume/hour after surgery increased the risk of developing pulmonary edema after surgery (hazard ratio: 1.03, 95% con dence interval [1.01-1.05], P = 0.013) and the risk of developing pulmonary edema with hypoxemia (hazard ratio: 1.04, 95% con dence interval [1.01-1.07], P = 0.024).

Conclusion
In this study, an increase in the transfusion of red blood cells per hour after surgery increased the risk of developing pulmonary edema after surgery. This increase can be considered a risk factor for the incidence of pulmonary edema.

Background
Massive bleeding during surgery is a life-threatening event in the operating room and increases postoperative mortality. [1] When massive bleeding occurs during surgery, When massive bleeding occurs during surgery, a relatively large volume of uid and blood is administered to compensate for the loss thereof, which could result in uid overload. Fluid overload with high hydrostatic pressures resulting in left ventricular dysfunction would be the most common consequence of postoperative pulmonary edema. [2] Massive transfusion after massive bleeding may cause coagulopathy, [3] acid-base abnormalities, [4] hypothermia, [5] and transfusion-related acute lung injury. [6] These complications of massive transfusions may also be related to pulmonary edema. [7][8][9][10] Intraoperative uid therapy is complex because it generally includes replacement of the maintenance requirement, existing uid de cits, and surgical wound losses including blood loss. Moreover, it is challenging to evaluate accurately the bleeding volume during surgery [11] and to perform transfusion safely. [12] After surgery, more uid and blood may be needed due to postoperative bleeding, or complications due to intraoperative massive transfusion. Therefore, administration of uid and transfusion therapy are more di cult after surgery. Moreover, currently, there is no clinical approach to safely administer uid and blood. Surgeons and anesthesiologists perform uid and transfusion therapy based on a combination of various clinical approximations. Perioperative uid and transfusion management in massive bleeding during surgery continues to be a daily challenge to surgeons and anesthesiologists.
In this study, we aimed to determine the intra-and postoperative uid and transfusion volume, which can be measured accurately. We also investigated the effect of the volume of uid and blood administered during and after surgery on the development of postoperative pulmonary edema in patients with intraoperative massive bleeding using a time-varying hazards analysis.

Data collection
This retrospective cohort study was approved by the Clinical Research Ethics Committee of Chuncheon Sacred Heart Hospital, Hallym University. All data were obtained from the clinical data warehouse of 5 hospitals at Hallym University Medical Center, between January 1, 2010, and December 31, 2019. The clinical data warehouse is a database of medical records, prescriptions, and test results; the data can be searched in multiple ways to identify patients, i.e., by prescription, examination type, and diagnosis. The time and results of patient tests can be extracted, in addition to their drug administration status, transfusion status, and overall medical records, in an unstructured text format.

Patients
Bleeding is divided into four classes according to the American College of Surgeons' Advanced Trauma Life Support classi cation. [13] Class 4 corresponds to a loss of > 40% of the circulating blood volume. In this study, estimated blood loss > 40% of average blood volume (males: body weight [kg] 75 ml; female: body weight [kg] 65 ml) [11] during surgery was de ned as massive bleeding; patients with massive bleeding during surgery were included. Patients meeting speci c criteria were excluded, which include: Patients < 18 years of age; Patients undergoing cardiac surgery; Patients who underwent a previous surgery within 7 days before the present surgery.
Patients with pulmonary edema or hypoxemia (PaO2/FiO 2 ≤ 300) before surgery Although no arterial blood gas (ABG) analysis was found, patients who did not receive oxygen therapy due to no respiratory symptoms after surgery and did not have pulmonary edema ndings on chest X-ray were considered to have absence of hypoxemia.

Primary and Secondary Outcomes
The primary outcome was the presence of pulmonary edema on postoperative chest X-ray. chest X-rays were used for consistency in the diagnosis of pulmonary edema, and all chest X-rays were evaluated by radiologists. The secondary outcome was pulmonary edema with hypoxemia after surgery. PaO2/FiO2 ≤ 300 was de ned as hypoxemia. The results with the least time interval between chest X-ray and ABG analysis were used. Patients who had no pulmonary edema on chest X-ray and did not receive oxygen therapy due to no respiratory symptoms were considered to have absence of pulmonary edema. Since postoperative pulmonary edema could occur up to 3 days after the operation, [12] it was con rmed whether postoperative pulmonary edema occurred until 72 hours after surgery.

Major variables
The major variables to be evaluated in this study were divided into intra-and postoperative variables. Intraoperative variables include the total amount of uid administered during surgery, the amount of uid administered per hour, the total amount of red blood cells administered, the amount of red blood cells administered per hour, the total amount of frozen fresh plasma (FFP) administered and the amount of FFP administered per hour.
Postoperative variables include the total amount of uid administered after surgery, the amount of uid administered per hour, the total amount of red blood cells administered, the amount of red blood cells administered per hour, the total amount of FFP administered and the amount of FFP administered per hour. All postoperative major variables were used as time-varying variables. The amount of uid and blood administered after surgery were measured based on the ABG analysis test time or the chest X-ray test time. In the analysis of pulmonary edema with hypoxemia, when both ABG analysis and chest X-ray were tested, the dose was measured based on the ABG analysis test time, and when there were no respiratory symptoms and only the chest X-ray test was performed, the amount was measured based on the chest X-ray test time. The administered uid and blood volume were expressed as a percentage of the patient's average blood volume. Each patient had 1-3 time-varying variables and observation periods depending upon the number of measurements according to the test frequency. The rst observation period was from the end of anesthesia to the time of the rst test; the second observation period was from the time of the rst test to the second test, and the third observation period was from the second test to the third test.
Postoperative variables included patient-controlled analgesia and whether or not creatinine increased by 0.3 compared to that before surgery. [15] Statistical Analysis Continuous variables were expressed as median and interquartile ranges due to skewness. Categorical variables were summarized as frequency and percentages. Continuous data were analyzed using the Mann-Whitney test, to compare patients with postoperative pulmonary edema and patients without postoperative pulmonary edema. Categorical data were analyzed using the chi-square test. Cox's timevarying hazards model was used to analyze the hazard ratio of the major variables for the occurrence of postoperative pulmonary edema within 72 hours after surgery.  Tables 1 (page 19) and 2, respectively. The median of the time interval between the chest X-ray and ABG analysis was 2.4 (interquartile range, 1.1-5.6) hours.
Hazard ratio of postoperative pulmonary edema with and without hypoxemia postoperatively within 72 hours The unadjusted and adjusted hazard ratio for major variables and kidney-related variables (preoperative glomerular ltration rate, intraoperative urine output ≤ 0.5 ml/kg/hour, postoperative creatinine increasing) of pulmonary edema with and without hypoxemia postoperatively within 72 hours are shown in Figure 1 and Figure 2, respectively. Postoperatively administered total uid (adjusted hazard ratio , 1.00; 95% con dence interval, 1.00-1.00: P < 0.001) and postoperative red blood cells per hour (adjusted hazard ratio , 1.03; 95% con dence interval, 1.01-1.05: P = 0.013) showed signi cant difference in hazard ratio in the unadjusted and adjusted analyses of postoperative pulmonary edema. Postoperative administered total uid (adjusted hazard ratio , 1.00; 95% con dence interval, 1.00-1.00: P = 0.005) and postoperative red blood cells per hour (adjusted hazard ratio , 1.04; 95% con dence interval, 1.01-1.07: P = 0.024) showed signi cant differences in hazard ratio in the unadjusted and adjusted analyses of postoperative pulmonary edema with hypoxemia. Intraoperative FFP per hour had a signi cant difference in hazard ratios only in postoperative pulmonary edema (adjusted hazard ratio, 1.03; 95% con dence interval, 1.00-1.06; P < 0.042). The other major variables did not show signi cant results. The data of adjusted hazard ratio of variables that are not shown in Figures 1 and 2 are shown in Table 3 (page 27) and 4 (page 28).

Discussion
Through time-varying Cox regression, it was shown that the total amount of uid administered after surgery and the amount of red blood cells administered per hour were associated with an increased risk of postoperative pulmonary edema with hypoxemia as well as postoperative pulmonary edema. However, the hazard ratio for the occurrence of postoperative pulmonary edema in the total amount of uid administered after surgery was so small that the clinical signi cance was not signi cant.
Our ndings show that in patients with massive intraoperative bleeding, the amount of red blood cells administered per hour after surgery increases the risk of developing postoperative pulmonary edema.
Exacerbation of pre-existing anemia, bleeding during surgery and repeated laboratory blood test are common causes of postoperative anemia. [17][18][19] If the anemia is severe or di cult to correct, more red blood cell transfusions may be required. Anemia and pulmonary edema are interrelated. [20][21][22] Existing severe anemia can often be associated with salt and water retention,[23-25] and hormonal and metabolic changes due to low hemoglobin levels can also cause direct myocardial toxicity, myocardial hypertrophy, and increase salt and water retention. [20,26] Heart failure can lead to anemia through a variety of mechanisms such as iron de ciency, in ammation, low erythropoietin level, medication, hemodilution, and medullar dysfunction . [27] The essential principle of uid balance in the body is that the amount of water lost should be equal to the amount of uid received.
[28] However, increased weight is common in patients undergoing surgery; this is typically a result of a positive uid balance. [33] Massive bleeding also causes uid imbalance, and massive volumes of uid and blood should be administered rapidly in such cases. Massive administration of blood products may cause transfusion-associated circulatory overload, [14] and uid overload can in turn cause postoperative pulmonary edema. [2,31,32,34] In this study, there were no variables related to both the occurrence risk of postoperative pulmonary edema and postoperative pulmonary edema with hypoxemia in the amounts of uid and blood administered during surgery. Perioperative uids and blood transfusions can cause hypothermia, [35][36][37] which can lead to impaired coagulation and acidosis,[38, 39] as well as pulmonary edema. [9] Administered uid and blood following bleeding can occur due to dilutional coagulopathy, platelet dysfunction, brinolysis, or hypo brinogenemia. And they are prone to occur when massive bleeding and large amounts of uids and blood are administered as compensation for bleeding. [ Because pulmonary edema is not visible on chest X-ray until the amount of lung uid increases by 30% relative to normal volume, [53] if there is no sudden increase of lung uid due to close monitoring during surgery, postoperative pulmonary edema may not be visible. However, in patients with massive bleeding, if su cient resuscitation is not achieved during surgery, organ failure may occur after surgery, [54] which may affect the development of pulmonary edema after surgery. [55][56][57] Even after surgery, it is recommended to frequently monitor the patient's volume condition using readily available clinical parameters such as vital signs, urine output, lung auscultation, weight change, and net uid retention calculations [58,59] but even doctors do not seem to monitor closely until lung edema is diagnosed. [32] In this study, an increase in FFP/hour during surgery increased the risk of postoperative pulmonary edema. While FFP is used in cases of clotting factor de ciency to prevent clotting disorders, the use of FFP puts the patients at risk for transfusion-related acute lung injury [6] and can cause circulatory overload in patients with kidney or cardiopulmonary failure. [60] The fact that an increase in FFP/hour during surgery is not associated with an increased risk of postoperative pulmonary edema with hypoxemia is a matter for further study.
Kidney function may also be closely related to the development of pulmonary edema. [57,61,62] Our study evaluated renal function with preoperative glomerular ltration rate, intraoperative oliguria, and increased postoperative creatinine, and it was found that increased postoperative creatinine increased the risk of postoperative pulmonary edema. The kidneys play an important role in many homeostasis mechanisms in the body and communicate with the lungs by regulating acid-base balance, increasing oxygen-carrying capacity through red blood cell production, and regulating blood pressure through the renin-angiotensin-aldosterone axis. However, as the acute kidney injury (AKI) progresses, these processes can become impaired. [57] If there is anemia, shock, or hemodilution during surgery, the risk of developing AKI increases after the surgery, [63,64] but these are more likely to occur in patients with massive bleeding during surgery. [65][66][67] During anesthesia, because the blood vessels expand, uid retention may occur, [68] and since the operation time is shorter than the follow-up period after surgery, it may be di cult to see kidney damage due to oliguria during surgery. [69] In this study, time-varying Cox regression was used to evaluate the hazard ratio. Time-varying covariance occurs when a given covariate changes over time during the follow-up period, a phenomenon commonly seen in clinical studies. [16,70] In patients with massive bleeding during surgery, the patient's condition after surgery may change in various ways, and according to the change in condition, several tests are performed and the results are measured. In addition, the uid and transfusion volume will be modi ed accordingly. In such cases where there are many changes after surgery, a time-varying model may be appropriate.

Limitations
A strength of this study was that a large number of patients with massive bleeding during surgery were included. However, because it is di cult to perform a randomized controlled trial of massive bleeding patients, this study used a retrospective design, which is subject to bias. pulmonary edema is either cardiogenic or non-cardiogenic, but we could not stratify patients based on pulmonary edema type.
Although the different types of pulmonary edema have different causes, it is di cult to distinguish them due to their similar clinical features. In addition, the causes can be complex in critically ill patients [55] and both types of postoperative pulmonary edema can occur in patients with massive bleeding during surgery.
[34, 45, 46] The gold standard for determining the cause of acute pulmonary edema is the insertion of a pulmonary artery catheter but, this is not routinely performed.
[71] Instead, in this study, we assessed pulmonary edema using a much less invasive method, the chest X-ray, and considered certain radiographic features that may help determine the cause of the edema. [55,72] In addition, hypoxemia was considered and analyzed to evaluate the objectivity and severity of pulmonary edema symptoms. Finally, this study did not include data on the output of body uids after surgery. Except for some serious patients admitted to the intensive care unit, no data on postoperative uid output were available for the follow-up period. Instead, we evaluated the assessment of kidney function related to urinary excretion with changes in creatinine levels.

Conclusions
This study provides information that the incidence of postoperative pulmonary edema is high in patients with massive bleeding during surgery, and an increase in the amount of transfusion of red blood cells after surgery present in the perioperative uid and blood may be a risk factor for the incidence of pulmonary edema and thus expands the results of previous studies. In patients with massive bleeding during surgery, close monitoring and rapid correction of factors that cause postoperative anemia and bleeding may be more important. Further prospective population-based studies are needed to explain the mechanism underlying the association between postoperative red blood cells transfusion and the risk of developing postoperative pulmonary edema in patients with intraoperative massive bleeding. Availability of data and materials

Abbreviations
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.         Figure 1 Unadjusted and adjusted hazard ratio for postoperative pulmonary edema occurrence FFP: fresh frozen plasma