Changes in central venous saturation after major surgery, and association with outcome

ScvO₂ data were collected from adult patients enrolled in the randomised study of post-operative goal-directed therapy (GDT) [16]. All patients were deemed to be at high risk of post-operative complications and were admitted to the intensive care unit (ICU) immediately after major surgery. This study was approved by the Local Research Ethics Committee of St George's Healthcare National Health Service Trust.

All patients had arterial and central venous catheters placed before the commencement of surgery. The central venous catheter was positioned with the tip within the superior vena cava immediately above the right atrium. This position was verified by chest radiograph and adjusted if necessary. The following parameters were monitored continuously from arrival in the ICU immediately after surgery and for the next 8 hours: electrocardiograph, pulse oximetry, invasive arterial pressure, central venous pressure and cardiac output. Arterial and central venous blood gas analyses were performed by intermittent blood sampling and co-oximetry (ABL 700; Radiometer, Copenhagen, Denmark) at baseline and hourly during the 8 hours after surgery. This equipment was calibrated each hour, and routine quality control checks were performed. Cardiac output was measured by lithium indicator dilution and pulse power analysis (LiDCO plus system; LiDCO Ltd., Cambridge, UK). P-POSSUM (Portsmouth Physiologic and Operative Severity Score for the enUmeration of Mortality and morbidity) and APACHE II (Acute Physiology and Chronic Health Evaluation II) scores were calculated at admission to the ICU [17, 18]. Complications and deaths occurring within 28 days of enrolment were included in the data analysis. Complications were prospectively defined, diagnosed by clinical staff and verified by a member of the research team. This process involved daily inspection of notes, radiological investigations, laboratory data and clinical assessment.

Protocols for cardiovascular management during the immediate post-operative period are provided in detail elsewhere [16]. Fluid challenges were guided by central venous pressure in 56 patients and by stroke volume in 61 patients. The latter group also received dopexamine if they did not achieve a DO₂I of 600 ml min^-1 m^-2 with fluid alone (GDT group). Once the 8-hour study period was complete, all patients received standard care for the remainder of their ICU and hospital stay. ScvO₂ data were not used to guide clinical management at any stage.

Data are presented as means ± SD where normally distributed, as medians (interquartile range) where not normally distributed or, for categorical variables, as a percentage of the group from which they were derived. Normality was tested with the Kolmogorov–Smirnov test. Categorical data were tested with Fisher's exact test. Continuous data were tested with the t test where normally distributed and the Mann–Whitney U test where not normally distributed. Trends in physiological parameters over time in the two groups were compared with repeated-measures analysis of variance with Tukey's correction for multiple comparisons.

Univariate analysis was performed to test association with complications and death. For data recorded hourly during the study period, the baseline values, lowest values and the mean over the 8-hour study period were tested. A multiple logistic regression model was used to identify independent risk factors for post-operative complications. A stepwise approach was used to enter new terms into the logistic regression model, where p < 0.05 was set as the limit for inclusion of new terms. Results of logistic regression are reported as adjusted odds ratios (ORs) with 95% confidence intervals. Receiver operator characteristic curves were constructed to identify optimal cutoff values for association with outcome. The optimum cutoff was defined as the value associated with the highest sum of sensitivity and specificity. Analysis was performed with GraphPad Prism version 4.0 for Windows (GraphPad Software, San Diego, CA, USA) and significance was set at p < 0.05.

Commonly measured physiological, biochemical and demographic variables are presented in Tables 1 and 2. Although derangements in CI, DO₂I and ScvO₂ were frequently observed, other parameters remained within the normal range or were only slightly abnormal. Univariate analysis identified five variables associated with post-operative complications. These were the lowest ScvO₂ value, the lowest DO₂I value, the lowest CI value, the P-POSSUM score and the use of GDT. After multivariate analysis, the lowest CI value (OR 0.58 (95% confidence interval 0.37 to 0.9); p = 0.018), the lowest ScvO₂ value (OR 0.94 (0.89 to 0.98); p = 0.007) and P-POSSUM score (OR 1.09 (1.02 to 1.15); p = 0.008) were independently associated with post-operative complications. The lowest DO₂I value and use of GDT were not independent predictors of outcome. The optimal value of ScvO₂ to discriminate between patients who did or did not develop complications was 64.4% (sensitivity 67%, specificity 56%). Univariate analysis identified no associations with mortality.

Patients were divided into two groups by using the optimal cutoff value for ScvO₂. Those in whom the lowest ScvO₂ value was 64.4% or below were defined as the low ScvO₂ group and those in whom the lowest value was above 64.4% were defined as the high ScvO₂ group (see Table 3). Trends in ScvO₂ and DO₂I are presented in Figures 1 and 2. During the first post-operative hour there was a significant decrease in ScvO₂ in both the high ScvO₂ group (79.8 ± 6.3% to 77.7 ± 5.8%; p = 0.016) and the low ScvO₂ group (74.6 ± 9.7% to 66.6 ± 10.3%; p < 0.0001). DO₂I and CI values did not change significantly during this time.