Peripheral measurements of venous oxygen saturation and lactate as a less invasive alternative for hemodynamic monitoring

The study was conducted as a prospective single-centre observational study. Patients who underwent elective cardiac surgery, from April 29, 2015 to May 16, 2015 at Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark, were included. Patients with a central and a peripheral venous cannula were eligible for the study. Exclusion criteria were age below 18 years and acute surgical patients.

The collection and processing of data were approved by the Danish Data and Protection Agency (j. no: 2012-58-0004). The regional ethics commission of the Copenhagen Region waived the need for informed consent as the study was appreciated as a quality-improvement study.

Surgery was performed through a median sternotomy with patients on cardiopulmonary bypass (CPB). Anaesthesia was induced with propofol 1–2 mg/kg i.v. and fentanyl 5 μg/kg i.v., and maintained with sevoflurane 0.5–3% and remifentanil infusion 0.3–.06 mcg/kg/hour. All patients were intubated and mechanically ventilated with 8 ml/kg, frequency 10–12/min, FiO2 0.6, PEEP 5–10 cmH2O, pCO2 4.5–6.5 kPa. Perioperative treatment of transfusions, fluid or drug infusion was given according to the National Guidelines and routine management at the department.

The CVC was placed after anaesthesia induction. Routinely a second peripheral cannula was inserted simultaneously with placement of the CVC. The peripheral intravenous site was determined by the treating physician with no input from the study investigators. Venous blood samples were obtained simultaneously from the central and peripheral venous cannula, in the latter after application of a tourniquet and with a maximum of 5 min between the two samples. The peripheral venous cannula site and tourniquet application time was registered. Any fluid or drug infusion was discontinued during blood sampling. Approximately 0.2 mL of blood was collected for analysis of gases after an adequate waste draw of 5–10 mL. Blood samples were obtained at pre-specified time-points: as soon as the CVC was inserted, upon arrival in the ICU and 3–4 h postoperatively. All blood samples were analysed on an ABL 800 FLEX radiometer. Results from the central venous blood gases were available to the treating physician.

A Bland-Altman plot was performed to describe the agreement between the two methods at all three time points. The bias and 95% limits of agreement (LoA) with confidence interval (CI) and the percentage error (PE) calculated according to Bland-Altman for all measurements [22]. The PE was calculated as 1.96 times SD of the bias divided by the mean value of ScvO₂ and SpvO₂. The results were categorized according to previous studies defining the agreement between two measurements using Bland-Altman plots [23, 24].

In a sub analysis the preoperative data were divided in to two subgroups (SpvO₂>/< 93%) to analyse the arterialisation of SpvO₂ observed in some patients. A four-quadrant plot was constructed to assess the trending ability.

Statistical analyses were performed using SPSS version 22.

The mean value of ScvO₂ and SpvO₂ was 80% (±7) and 93 (±8) preoperatively, 72% (±7) and 83% (±13) on arrival to the ICU and 71% (±7.5) and 87% (±10) at 3–4 h postoperatively, respectively. Venous oxygen saturation was on average higher when obtained from a dorsal hand vein compared to cubital samples (Table 1).

The Bland-Altman plots for venous oxygen saturation are presented in Fig. 1a–c. Bias was 13.37% in the preoperative measurements (95% CI: 11.52–15.22, LoA: ±19.10), 11.29% at arrival to the ICU (95% CI: 8.81–13.77, LoA: ±25.10) and 16.49% at 3–4 h postoperatively (95% CI: 14.16–18.82, LoA: ±21.20). Preoperatively at baseline, 50% of the blood samples had a SpvO₂ > 93%, indicating arterialisation of blood. For this reason, a sub analysis with the Bland-Altman plot was performed in two separate groups: (SpvO₂ >/< 93%). Bias was 17.20% (95% CI: 15.80–18.59, LoA: ±11.90) and 4.94% (95% CI: 1.09–8.79, LoA: ±21.60) in the high and the low preoperative group, respectively. The percentage error was 22.08% preoperatively, 32.39% at arrival to the ICU group and 26.83% at 3–4 h postoperatively. The linear regression line was R² =0.12.

The trending ability of ScvO₂ compared with SpvO₂ in all measurements was investigated in a four-quadrant plot as demonstrated in Fig. 2. The association between central and peripheral venous oxygen saturation is presented in a scatter plot with a line of identity (Fig. 4a). The plot shows an 89% concordance with no zone of exclusion and 77% concordance including the zone of exclusion.

The mean value of central and peripheral lactate was 0.87 (±0.28) and 1.00 (±0.35) preoperatively, 1.71 (±1.15) and 1.89 (±1.22) on arrival to the ICU and 1.85 (±1.08) and 2.12 (±1.41) at 3–4 h postoperatively, respectively.

The Bland-Altman plots for lactate preoperatively are presented in Fig. 3a–c. Bias was 0.14 mmol/L preoperatively (95% CI: 0.11–0.17, LoA: ±0.30), 0.16 mmol/L at arrival to ICU (95% CI: 0.09–0.23, LoA: ±0.70) and 0.23 mmol/L at 3–4 h postoperatively (95% CI: 0.11–0.35, LoA: ±0.50). The percentage error was 32.08% preoperatively, 38.88% at arrival to the ICU group and 25.18% at 3–4 h postoperatively. The association between central and peripheral lactate is presented in a scatter plot with a line of identity (Fig. 4b).

Our results showed an acceptable LoA and a moderate trending ability, indicating an acceptable agreement between SpvO₂ and ScvO₂. The precision of the individual methods was not determined in this study and the exact value of an acceptable LOA of agreement could not be calculated. However, a percentage error of 22% to 32% is rated to be within the clinically acceptable [23]. Measurements of peripheral venous oxygen saturation and lactate could be valuable in different populations such as emergency or trauma settings, avoiding unnecessary and time consuming application of a central venous or a pulmonary artery catheter, and in order to quickly identify patient at risk and hemodynamics. Routinely all patients get a central and peripheral line prior to cardiac surgery at our department, which allowed us to investigate the feasibility of SpvO₂ in this patient population. Additional studies are required to investigate whether these observations are reproducible in a different setting. The present study is limited by the fact that only a few patients demonstrated low cardiac output postoperatively and therefore, the clinical implication of SpvO₂ in patient with low cardiac output is unknown. Surprisingly, a high SpvO₂ was seen in many patients at baseline measurements. A simultaneously high ScvO₂ indicated arterialisation in these patients. The risk of arterialisation at different ScvO₂ values is unknown and should be further investigated. Other limitations to this study were differences in the peripheral intravenous site, which may have altered the results. Tourniquet application in some patients may have caused hemolysis in the peripheral blood samples. However, very few patients had permanence use of a tourniquet during peripheral blood sampling. Venous blood samples were obtained simultaneously from a central or a peripheral venous catheter however, delays may have occurred due to waste drawn.