Using clinical parameters to guide fluid therapy in high-risk thoracic surgery. A retrospective, observational study

The development of ARDS as characterised in the Berlin Definition from 2011 [3] is a well-known complication to pulmonary resection, and to pneumonectomy in particular. In 1984, Zeldin [7] proposed that excessive fluid administration was the cause of pulmonary oedema after pneumonectomy. Since then several mechanisms have been proposed to take part in the development of what may be defined as ARDS after pneumonectomy, and excessive fluid administration might just be one of several contributory factors [4].

In 2002, Møller et al. [8] showed that an excess fluid balance of more than 4 l during surgery was associated with a higher risk of postoperative complications after pneumonectomy and in 2010, Marret et al. [9] found that liberal fluid administration during surgery was a risk factor for major complications after pneumonectomy. Furthermore, Suehiro et al. [10] found that high fluid infusion volumes during pneumonectomy were a risk factor for developing postoperative right-sided heart failure. However, even though fluid restriction seems to be an important factor in reducing complications after pneumonectomy, it may compromise cardiac output and ultimately lead to inadequate oxygen delivery. As recently discussed [11], the use of haemodynamic monitoring by devices such as transesophageal Doppler, PiCCO etc. may prove useful for guiding perioperative fluid therapy in pulmonary resection surgery, but still further research is needed in this field. However, goal-directed therapy using measurements of cardiac output alone, often leads to higher volumes of administered fluid compared to conventional therapy [12], and therefore the debate continues as to the optimal dose of fluid and which haemodynamic goal and/or vasopressor are best [13].

According to the Fick equation mixed venous oxygen saturation (SvO2) is related to the fraction of delivered oxygen that is consumed by the tissues, and a reduced SvO2 is thought to reflect elevated oxygen extraction, compensating for a reduced cardiac output [14]. Since ScvO2 has been shown to correlate with SvO2 [15], the former was used to avoid insertion of a pulmonary artery catheter. In sepsis treatment, a ScvO2 > 70 % is widely used and it has been suggested that normovolemia in the supine position is reached when fluid administration results in a ScvO2 higher than 65 % [16]. Furthermore, it has been suggested that perioperative mean values of ScvO2 < 71–73 % are correlated to an increased risk of postoperative complications - predominantly infections - after high-risk abdominal surgery [17, 18]. Consequently, a ScvO2 of 60 % as interventional threshold could be considered to be a rather low. However, in pulmonary resection surgery, the main concern is still the development of acute lung injury and ARDS [11], which is why the main purpose of implementing BGDT was to avoid hypervolemia by restricting fluid administration without compromising the circulation and oxygen delivery. Thus, due to the possible hazardous consequences of hypervolemia in patients undergoing EPP a relatively low interventional set point for ScvO2 was chosen. Also, it has been suggested that the optimal interventional threshold values of ScvO2 are dependent on the study population [17] which is why future prospective studies are needed to establish optimal target values of ScvO2 adapted to specific surgical populations.

Attempting to minimize the total expansion of body interstitial fluids, we wished to restrict the use of crystalloids but at the same time avoiding the use of blood products for volume substitution. Therefore, we changed the recommended colloid of choice from dextran to hydroxyethyl starch, allowing administration of a higher dose in 24 h. We acknowledge that the use of synthetic colloids is controversial due to results from multicentre studies on sepsis treatment [19, 20], but similar studies in surgical patients are still anticipated. Beside the change of colloid, the new guidelines also clearly emphasised the restrictive use of blood products which may have contributed to the observed tendency towards reduced volumes of administered blood products and the reduced proportion of patients, who received blood transfusion in the after group. Also, the total volume of intraoperative fluids were reduced (Table 3) resulting in less hemodilution.

Radiological signs of pulmonary stasis were present on chest X-rays from the first POD in one patient from the before group and in three patients from the after group, but none of the four patients had clinical symptoms of pulmonary stasis or oedema, and all were discharged from the ICU on POD 1. The implementation of BGDT did not seem to increase the incidence of kidney injury according to the creatinine criteria of the RIFLE-classification [21] neither did it increase the use of inotropic agents in the ICU (Table 3).

A limitation to this retrospective, observational study is that the impact of the changes to standard therapy, were compared to a group of historical, non-matched patients and therefore the results could be dependent on non-measured variables among patients. However, this group of patients is a highly selected group without significant comorbidity and the data were comparable regarding basic characteristics, suggesting an acceptable matching between groups. Furthermore, the main part of the data was collected from electronic databases, which might result in a higher level of data validity and completeness. Another limitation is that several changes were applied at the same time, which prevents us from drawing conclusions on the impact of the individual changes. Also, our results could be influenced by variables of clinical, technical, or logistic origin that were not controlled during the study period. Finally, the small sample size and hereby potential lack of statistically power precludes us from drawing conclusions on risk and benefits and therefore our results should be considered hypothesis generating, serving for designing further investigations.