Association Between Early Central Venous Pressure Measurement and Mortality in Patients With Sepsis: A Retrospective Study

Background: It remains controversial to abandon central venous pressure (CVP) in the monitoring of uid resuscitation in sepsis patients. We hypothesized that early CVP measurement was associated with decreased mortality in critically ill patients with sepsis. Methods: Critically ill patients with sepsis were identied from the Medical Information Mart for Intensive Care (MIMIC)-III database. Patients were divided into two groups base on whether or not they had a CVP measurement within the rst 6 hours of ICU stay. The primary endpoint was 28-day mortality. Patients were further divided into four subgroups base on the time when the rst CVP measurement was obtained: within 3 hours, between 3 and 6 hours, between 6 and 12 hours, and not measured within 12 hours. Kaplan-Meier survival analysis and Cox regression model were used for univariate and multivariate analyses of survival, respectively. RESULTS: A total of 4733 sepsis patients were included. The 28-day mortality was signicantly lower in the CVP measured group than in the control group (34.2 vs. 40.7, p < 0.001). A “U”-shaped relationship between initial CVP and mortality was identied. With patients without CVP measured within 12 hours serving as the reference subgroup, timely CVP measurement was associated with decreased 28-day mortality before and after adjusting for confounding variables. CONCLUSIONS: Early CVP measurement is associated with decreased 28-day mortality in patients with sepsis. CVP should be considered as a valuable and easily accessible safety parameter during (early) uid resuscitation. mean standard deviation median range) continuous variables, categorical


Background
Sepsis and septic shock remain the main causes of admission to the intensive care unit (ICU) and death in critically ill patients [1]. During this rst phase of treatment ("the salvage phase" [2]), one of the main therapies is uid administration to restore blood pressure and cardiac output. Previous Surviving Sepsis Campaign (SSC) guidelines have recommended administering uid to target central venous pressure (CVP) level [3], yet it is no longer recommended by the latest updated SSC guidelines [4]. It remains controversial to abandon CVP in the monitoring of uid resuscitation in sepsis patients [5,6]. CVP is often misinterpreted as an indicator of intravascular volume status. From the (patho)physiological point of view, CVP is determined by the interaction of cardiac function and venous return function, as suggested by Guyton et al. [7], and therefore does not merely re ect the intravascular volume status. Small changes in CVP along with increases in cardiac output (CO) indicate uid responsiveness, which would thus lead to uid administration in sepsis patients with evidence of tissue hypoperfusion. In contrast, signi cant increases in CVP with little changes in CO suggest poor tolerance to uids, in which case patients are at high risk of uid overload and may bene t little from uids.
It is true that using CVP values to guide uid resuscitation is far from perfect. However, CVP provides a useful indication of how well the heart interacts with venous return. To a given high CVP value, it could attribute to either cardiac dysfunction or uid overload, or both, if intra-thoracic hypertension caused by other factors is excluded. CVP can thus serve as a safety variable for when to stop or not to give additional uids. This is important since sepsis patients admitted to ICU is very likely to receive a large amount of uid before ICU admission. Low CVP, on the other hand, can be helpful for detecting inadequate uid resuscitation but should be used with caution and together with additional monitoring. We hypothesized that early CVP measurement in the ICU was associated with decreased mortality in critically ill patients with sepsis. Data of patients with sepsis were retrospectively collected, and the association between early CVP measurement and mortality was explored.

Database
The data of the present study were collected from the Multiparameter Intelligent Monitoring in Intensive Care Database III (MIMIC III) [8]. In brief, MIMIC-III database is maintained by the Laboratory for Computational Physiology at the Massachusetts Institute of Technology. It contains de-identi ed health-

Data extraction
PostgreSQL tools V.1.09 was used for data extraction. The following data were extracted by using Structured Query Language (SQL): age, gender, co-morbidities, the survival time, length of hospital stay, and length of ICU stay, sequential organ failure assessment (SOFA) score, vital signs, rst-day lab results, use of vasopressors, daily uid input, uid balance, urine output, need for renal replacement therapy and mechanical ventilation, the present of sepsis, and the initial time and value of the rst CVP measurement in the ICU.

Population and endpoint
All patients with sepsis were included. Patients younger than 18 years old were excluded. Patients were divided into two groups base on whether or not they had a CVP measurement within the rst 6 hours of ICU stay (i.e., the CVP measured group and the control group). The primary endpoint was 28-day mortality.

Statistical analysis
Continuous variables were presented as mean with standard deviation or median with interquartile ranges. Student's t-test or Wilcoxon rank-sum test were used as appropriate. Categorical variables were presented as counts (percentages) and compared using the chi-square test. We used a locally weighted smoothing (Lowess Smoothing) technique to explore the crude relationship between initial CVP values and mortality. Patients were further divided into four subgroups base on the time when the rst CVP measurement was obtained: within 3 hours, between 3 and 6 hours, between 6 and 12 hours, and not measured within 12 hours, and the Kaplan-Meier survival analysis was performed as a univariate analysis of survival, tested by the Log-Rank test. Cox regression model was used to estimate the relationships of the timeliness of CVP measurement and mortality adjusting for confounding variables (with a p-value < 0.05 in univariate analysis or potential confounders judged by clinical expertise).

Results
A total of 4733 sepsis patients were included, where 1673 patients had a CVP measurement within the rst 6 hours of ICU stay, and another 3060 patients were allocated into the control group. Baseline patient characteristics are summarized in Table 1. It is worth mentioning that patients in the CVP measured group had higher initial lactate acid level, higher SOFA score, but a lower rate of underlying renal failure (Table 1). CVP central venous pressure, SOFA sequential organ failure assessment The 28-day mortality was signi cantly lower in the CVP measured group than in the control group (34.2 vs. 40.7, p < 0.001, Table 2). Compare to the control group, the hospital length of stay was signi cantly shorter in the CVP measured group, but the ICU length of stay was signi cantly longer. More patients needed mechanical ventilation in the CVP measured group, but there was no difference in the duration of mechanical ventilation. Similarly, more patients needed vasopressors in the CVP measured group, but no difference in the duration of vasopressor use was observed. No difference in the occurrence of acute kidney injury (AKI), need for renal replacement therapy, or urine output was observed ( Table 2). A "U"-shaped relationship between initial CVP and mortality was found (Fig. 2). The 28-day survival was signi cantly different among patients with CVP measured at different time points (Fig. 3). Compare to patients without CVP measured within 12 hours, patients with CVP measured within 3 hours had signi cantly lower mortality, but no signi cant difference was observed in the other subgroups (adjusted p < 0.001, = 0.053, and > 0.999, respectively; Fig. 3).
With patients without CVP measured within 12 hours serving as the reference group, timely CVP measurement was associated with decreased 28-day mortality before adjusting for confounding variables. After adjusting age, gender, SOFA score, initial lactate, renal failure, and the rst-day lab results, including max blood glucose, minimum platelet and max white blood cell count, the association between timely CVP measurement and decreased 28-day mortality was still observed (Table 3). Patients were divided into four subgroups base on the time when the rst CVP measurement was obtained: within 3 hours, between 3 and 6 hours, between 6 and 12 hours, and not measured within 12 hours. Two COX proportional hazards models were built with the CVP not measured within 12 hours subgroup as the reference group.
CVP central venous pressure, CI con dence interval, HR hazard ratio, SOFA sequential organ failure assessment, WBC white blood cell count

Discussion
The main ndings of the present study were: 1) the 28-day mortality was signi cantly lower in patients with CVP measured within 6 hours than in those who did not; 2) the improvement of 28-day mortality was associated with the time when the CVP measurement was established.
The results of three recent multi-center randomized controlled trials (ProCESS, ARISE and ProMISe) have led to the abandonment of the "early goal-directed therapy" (EGDT) strategy [9,10]. CVP is no longer recommended to guide uid resuscitation by the most recent SSC guidelines [4]. Instead, dynamic measures of assessing uid responsiveness, encompass passive leg raises, uid challenges against stroke volume measurements, or the variations in systolic pressure, pulse pressure, or stroke volume to changes in intrathoracic pressure induced by mechanical ventilation, are recommended. However, new concerns regarding the concept of uid responsiveness have been raised: "volume responsive, but does the patient need volume?" [11] What is the signi cance of CVP measurement in this context? CVP is determined by the complex interplay of the preload, afterload, compliance, and contractility of the right ventricle, venous tone, volume status, abdominal and intrathoracic pressure, and many other factors [12]. For simplicity, it can be interpreted as that measured CVP is determined by the interaction of the cardiac and venous return functions [13].
Indeed, a single static measurement of CVP does not predict blood volume or uid responsiveness. However, this statement misses the value of CVP as a readily available indicator that mirrors the complex interaction between the heart and the blood returning from circulation. In other words, CVP is more likely a "dependent variable" in this interplay rather than an "independent variable." The "independent variables," on the other hand, are intravascular volume and cardiac function, which are manipulated in daily clinical practice (for instance, uid administration for intravascular volume and inotropic agents for cardiac function). In this point of view, CVP does not indicate volume status, and in isolation, it is not a strong predictor of uid responsiveness. But an abnormal CVP value must attribute to either abnormal intravascular volume or abnormal cardiac function, or both, and the choice of further (advanced) assessments can be made by the measurement of CVP (e.g., echocardiography, inferior vena cava diameter, CO measurement via thermal dilution, etc.). Together with the further evaluations, the decision whether to continue or stop uid administration or even to achieve a negative uid balance (i.e., "reverse uid resuscitation"), can be made.
In the present study, our data suggested a "U" shape relationship between initial CVP and 28-day mortality. As discussed above, an extremely low CVP value may due to hypovolemia or a hyperdynamic heart in the setting of sepsis. Patients are at higher risk of mortality if hypovolemia is inadequately resuscitated [4]. The association between elevated CVP and increased mortality has been found by numerous studies [14][15][16][17][18][19]. Besides, since CVP serves as the backpressure of venous return, upstream capillary pressures must be even higher in the setting of increased CVP to maintain an intact venous return. Therefore, clinicians should not only further look into cardiac function or intravascular volume, but also the possible impact of venous congestion on upstream organs. The association between elevated CVP and kidney dysfunction has also been found in critically ill patients [18,20,21]. In the perspective of venous return function, under the premise of maintaining adequate tissue perfusion, CVP should be maintained as low as possible to allow an adequate venous return, and thus to increase CO (in a steady condition, CO must equal venous return) and reduce tissue edema.
Criticisms have been raised since the current SSC guidelines recommend intravenous infusing at least 30 mL/kg of crystalloids within the rst three hours of resuscitation [5]. A xed, large volume of uid infusion could result in uid overload. As previously discussed, a high CVP indicates either uid overload or the current cardiac function is incapable of matching the increased venous return. Either way, the uid infusion should be discontinued. Early CVP monitoring may be helpful in reducing the risk of uid overload. In this setting, CVP can serve as a safe limit rather than a target of uid resuscitation [22]. Our data suggested that early CVP measurement was associated with decreased 28-day mortality, which can be partially explained in this way.
Higher baseline SOFA score and initial lactate were observed in patients who received CVP measurement within 6 hours, suggested a more severe illness of the patients. Interestingly, these sicker patients had better survival than the controls. After adjusting the confounding variables in multivariate analysis, we found that early CVP measurement was the independent factor associated with decreased mortality. However, it should be emphasized that it is not the monitoring, but the adjustment of interventions given by the caregivers based on the results of the monitoring improves patients' outcomes. Although a causeeffect relationship cannot be established in this retrospective study, given the nature of cheap, easily obtainable, and repeatable, early CVP measurement is still a valuable aid for diagnosis and hemodynamic management when properly understood, especially in preventing uid over-resuscitation.
This study has several limitations. First, the study was limited by the nature of retrospective design and the source of data used. For this reason, no cause-effect relationships could be established. Second, although some confounding variables were adjusted in multivariate analysis, con rming that early CVP measurement was associated with poor outcomes, the results may be affected by other confounding variables related to organ damage or mortality. Additional prospective studies are needed to investigate the association between CVP monitoring and treatment adjustments and the related outcomes to these adjustments.

Conclusions
Early CVP measurement is associated with decreased 28-day mortality in patients with sepsis. CVP should be considered as a valuable and easily accessible safety parameter during (early) uid resuscitation.
Abbreviations AKI acute kidney injury; CO = cardiac output; CVP = central venous pressure; EGDT = early goal-directed therapy; HR = hazard ratio; ICU = intensive care unit; MIMIC = Medical Information Mart for Intensive Care; SOFA = sequential organ failure assessment; SQL = Structured Query Language; SSC = Surviving Sepsis Campaign; WBC = white blood cell count Declarations Ethics approval and consent to participate The institutional review boards of the Massachusetts Institute of Technology (Cambridge, MA) and Beth Israel Deaconess Medical Center (Boston, MA) approved the establishment of the database. Therefore, consent was obtained for the original data collection and waived for the present study.

Consent for publication
Not applicable Availability of data and material The datasets used and analyzed during the present study are available from the corresponding author on reasonable request.

Figure 1
Flowchart showing a step-by-step selection of patients included in the study CVP central venous pressure, ICU intensive care unit.