Does a conservative fluid management strategy in the perioperative management of lung resection patients reduce the risk of acute lung injury?

Abstract

A best evidence topic in thoracic surgery was written according to a structured protocol. The question addressed was whether a conservative fluid management strategy in the perioperative management of lung resection patients is associated with a reduced incidence of postoperative acute lung injury (PALI) and/or acute respiratory distress syndrome (ARDS) in the recovery period. Sixty-seven papers were found using the reported search, of which 13 level III and 1 level IV evidence studies represented the best evidence to answer the question. Two retrospective case-control studies demonstrated a direct association between liberal fluid intake and the incidence of PALI/ARDS following lung resection on multivariate analysis (MVA) with odds ratios (ORs) of 1.42 (95% CI 1.09–4.32, P = 0.011) and 2.91 (1.9–7.4, P = 0.001). In non-PALI/ARDS cases, the mean intraoperative fluid infusion volume was significantly less [1.22 l (1.17–1.26) vs 1.68 l (1.46–1.9) P = 0.005], the fluid balance over the first 24 postoperative hours was significantly less [1.52 l positive (1.44–1.60) vs 2.0 l positive (1.6–2.4) P = 0.026] and cumulated intra- and postoperative fluid infusion was significantly less [2.0 ml/kg/h (1.7–2.3) vs 2.6 ml/kg/h (2.3–2.9) P = 0.003]. These data show that the difference between fluid regimes associated with an increased incidence of PALI/ARDS (i.e. ‘liberal’) and those which are not (i.e. ‘conservative’) is narrow but significant. However, this does not prove a causative role for liberal fluid in the multifactorial development of PALI/ARDS. On this best evidence, we recommend intra- and postoperative maintenance fluid to be administered at 1–2 ml/kg/h and that a positive fluid balance of 1.5 l should not be exceeded in the perioperative period with caution being exercised with regard to the adequacy of oxygen delivery. If the fluid balance exceeds this threshold, a high index of suspicion for PALI/ARDS should be adopted and escalation of the level of care should be considered. If a patient develops signs of hypoperfusion after these thresholds are exceeded, inotropic/vasopressor support should be considered.

INTRODUCTION

A best evidence topic was constructed according to a structured protocol as described in ICVTS [1].

THREE-PART QUESTION

In [perioperative lung resection patients] is [a conservative fluid management strategy] superior to [a liberal fluid management strategy] in terms of [reduced incidence of postoperative acute lung injury/acute respiratory distress syndrome]?

CLINICAL SCENARIO

A 61-year old man underwent an open left lower lobectomy. He received 2.1 l of intravenous crystalloid intraoperatively and in the first 24 h postoperatively, he received fluids at 125 ml/h. He weighed 85 kg. Eight hours postoperatively, his urine output was 10 ml/h for two consecutive hours. His fluid balance was 2.3 l positive for the last 24 h. You consider whether to administer a fluid challenge or to commence vasopressor/inotropic support mindful of the risk of postoperative acute lung injury (PALI)/acute respiratory distress syndrome (ARDS).

SEARCH STRATEGY

MEDLINE was searched from 1950 to October 2011, and EMBASE was searched from 1980 to October 2011 using the OVID interface. The search terms were: [acute lung injury.mp OR acute respiratory distress syndrome.mp OR pulmonary oedema.mp] AND [lung resection.mp OR pneumonectomy.mp OR lobectomy.mp] AND [intravenous fluid.mp OR fluid balance.mp]. The Cochrane library was also searched using the same terms.

SEARCH OUTCOME

Sixty-seven papers were found using the reported search. Fourteen papers were selected as giving the best evidence (Table 1).

Table 1:

Relevant papers

Author, date and country Study type (level of evidence)	Patient group	Outcomes	Key results	Comments
Licker et al . (2003) Anesth Analg, Switzerland, USA [3] Prospective case-control study (level III)	n = 879 Lung resections (all types) for NSCLC 1991–2002 Fluid regimen: 1 ml/kg/h in addition to replacement of losses intra- and postoperatively	PALI/ARDS: 37 (4.2%) (Using consensus criteria [4])	UVA: Cumulated intra- and postoperative crystalloid fluid infusion ml/kg/h (mean: 95% CI): PALI/ARDS group; 2.6 ml/kg/h (2.3–2.9) Non-PALI/ARDS group; 2.0 ml/kg/h (1.7–2.3) P = 0.003 MVA: Fluid infusion (cumulated intra- and postoperative) is associated with PALI/ARDS by an OR of 2.91 (95% CI: 1.9–7.4) P = 0.001	Increasing fluid infusion is shown to be associated with PALI/ARDS by UVA and MVA Factors included in the MVA: Chronic alcohol excess Pneumonectomy Ventilatory hyperpressure Fluid infused
Licker et al. (2009) Critical Care, Switzerland, USA [2] Retrospective case-control study (level III)	n = 1091 Lung resection (all types) for primary lung cancer 1997–2008 Fluid regimen: Intraoperative crystalloid: 2–4 ml/kg/h in addition to replacement of losses	Protected lung ventilation strategy group (2003–2008) n = 558 PALI/ARDS: 5 (0.9%) Standard ventilator strategy group (1997–2003) n = 533 PALI/ARDS: 19 (3.7%) (Using consensus criteria [4]) Both cohorts had similar baseline characteristics	UVA: OR: 1.33 (95% CI: 1.02–5.08) per 1 ml/kg/h increase in intra- and postoperative fluid administration P = 0.032 MVA: OR: 1.42 (95% CI: 1.09–4.32) per 1 ml/kg/h increase in intra- and postoperative fluid administration P = 0.011	Increasing fluid infusion is shown to be associated with PALI/ARDS by UVA and MVA Factors included in the MVA: Chronic alcohol excess Chemoradiotherapy Advanced TNM stage Pneumonectomy Fluid infused
Blank et al. (2011) Ann Thorac Surg, USA [6] Retrospective case-control study (level III)	n = 129 Pneumonectomy 1997–2008 Fluid regimen: intraoperative fluid management of 1 ml/kg/h titrated to a mean arterial pressure of 75% of preoperative average	All respiratory complications: 27 (21%) PALI/ARDS: 9 (7%) (Using consensus criteria [4])	UVA: All respiratory complications group: total intraoperative fluid; 2.7 l (95% CI: 2.0–4.0) No respiratory complications group; 1.8 l (1.5–2.5) P < 0.001 MVA: No significant association	Fluid is assessed against a composite outcome of all respiratory complications, not individually against PALI/ARDS Small cohort of PALI/ARDS cases Factors included in MVA: Blood product usage Total fluids per litre Surgical indication (benign vs malignant) Pneumonectomy type Duration of anaesthesia ASA status
Alam et al. (2007) Ann Thoracic Surg, USA [5] Retrospective case-control study (level III)	n = 152 Lung resection (all types) for primary lung cancer 2001–2004 Fluid regimen: Intraoperative fluid restriction was used for all patients	Primary lung injury: 76 (5.3%) Primary lung injury = clinical diagnosis of PALI/ARDS without available arterial blood-gas data PALI/ARDS: 44 (3.1%) (Using consensus criteria [4])	UVA: Primary lung injury group (n = 76): Mean perioperative fluids infused; 2.75 l (95% CI: 1.35–5) Control group (n = 76): 2.5 l (1.4–4.5) OR 1.2 (95% CI: 1.0–1.4) P = 0.05 MVA: For every 500 ml increase in perioperative fluid administration: OR 1.2 (95% CI: 1–1.4) for postoperative primary lung injury P = 0.02	Evidence on MVA that increasing perioperative fluid administration is associated with ‘primary lung injury’ Absence of arterial blood gas data on 32 patients means that the true incidence of PALI/ARDS according to [3] guidelines may be over- or under-estimated. Analysis of fluid volume against confirmed cases of PALI/ARDS (44) was not performed Group matching: Smoking status Extended resection? Sex pT, pN, pM status
Marret et al. (2010) Interact CardioVasc Thorac Surg, France [7] Retrospective case-control study (level III)	n = 129 Pneumonectomy 2000–2005 Fluid regimen: Not defined	PALI/ARDS: 9 (7%) (Using consensus criteria [4])	UVA: Total intra- and postoperative fluid infusion (l): Patients with all major complications (n = 55); 3.8 l ± 1.5 (mean ± SD) Patients without major complications (n = 74); 2.5 l ± 1.3 (mean ± SD) OR: 1.91 (95% CI: 1.47–2.83) P < 0.0001 MVA: OR 1.96 per litre increase (95% CI: 1.45–3.16) P < 0.0001	Total fluid infusion was assessed against all major complications rather than PALI/ARDS individually on UVA and MVA Small sample size (n = 9) Factors included in MVA: ASA class Age Smoking history COPD Haemoglobin Neutrophil count Urea and creatinine Total fluid infusion (l) Peak pressure (cm water) Tidal volume (ml/kg PBW) Transfusion (%) Operating time (min)
Fernandez-Perez et al. (2006) Anesthesiology, USA [8] Retrospective case-control study (level III)	n = 170 Pneumonectomy 1999–2003 Fluid regimen: Not defined	PALI/ARDS: 15 (9%) (Using consensus criteria [4]) Respiratory failure of other aetiologies: 15 (9%) Total incidence of respiratory failure: 30 (18%)	UVA: Respiratory failure of all causes (30): Median intraoperative fluid volume infused: 2.2 l (inter-quartile range: 1.4–3.7) Patients without respiratory failure (140): Median intraoperative fluid volume infused 1.3 l (inter-quartile range: 0.9–2.7) P < 0.001 MVA: Per litre of fluid infused intraoperatively: OR 1.34 (95% CI: 0.83–2.09) P = 0.201 (not significant)	Fluid input was not directly assessed against cases of PALI/ARDS No effect of increased fluid was found on MVA Factors included in MVA: Intraoperative VT Preoperative FVC Fluid Intraoperative VT x fluid
Parquin et al. (1996) Eur J Cardiothoracic Surg, France [11] Retrospective case-control study (level III)	n = 146 Pneumonectomy 1992–1992 Fluid regimen: Not defined	Pulmonary oedema within first postoperative week: 22 (15%) Severe pulmonary oedema within first postoperative week: 5 (3.4%) Pulmonary oedema criteria: Tachypnoea Crackles Edematous Sputum Diffuse infiltrates on chest radiographs	UVA: Total intraoperative fluid load ≥ 2 l Pulmonary oedema within first postoperative week (n = 22): 10 (45%) Control group (n = 124): 25 (20%). P < 0.01 MVA: Total intraoperative fluid load ≥2 l is associated with pulmonary oedema of all severities (P = 0.02)	No arterial blood gas data Consensus criteria [4] are not used Data confounded by mild pulmonary oedema Impossible to directly compare this study to others Factors included in MVA: Prior radiotherapy Total fluid load ≥2 l Remaining lung perfusion ≤55%
Moller et al. (2002) Eur Resp J, Denmark [9] Retrospective case-control study (level III)	n = 107 Pneumonectomy 4-year period Fluid regimen: Normal saline 0.9% was administered at 5 ml/kg/h intraoperatively	Pulmonary complications (all): 19 (7.8%) No measurement of incidence of PALI/ARDS	UVA: >4 l during anaesthesia (n = 13) 38.5% developed pulmonary complications <4 l during anaesthesia (n = 92) 15.2% developed pulmonary complications P < 0.05 >4 l fluid balance group associated with higher mortality (30.8 vs 7.6%, P < 0.05) MVA: Intraoperative fluid balance >4 l OR 4.48 (95% CI: 1.05–19.02) P = 0.042 No analysis of mortality on MVA	Consensus criteria [4] are not used Fluid balance/excess is not specifically analysed against PALI/ARDS Rationale for large volumes of intraoperative fluid administration not given Possibility data is confounded by blood loss Factors included in the MVA: Age >70 Pre-op heart disease BMI <17 or >25 Pneumonectomy (right) Anaesthesia >360 min Fluid excess >+4 l
Bernard et al. (2001) J Thorac Cardiovasc Surg, USA [10] Retrospective cohort study (level III)	n = 639 Pneumonectomy 1985–1998 Fluid regimen: Not defined	Post-pneumonectomy pulmonary oedema (PPPE): 7 (1.1%)	UVA: IV fluid variables associated with cardiopulmonary complications (all causes): IV crystalloid first 24 h (including intraoperative fluids) (P = 0.01) IV crystalloid first 12 h (including intraoperative fluids) (P = 0.01) IV fluid variables associated with mortality: IV crystalloid first 12 h (P = 0.01) MVA: No significant associations with IV crystalloid	PPPE is not defined Consensus criteria [4] are not used Primary data relating to fluid administration are unpublished Study is not controlled Fluids are not directly assessed against PALI/ARDS Factors included in MVA: Increased age Cardiovascular disease Muscle reinforcement of bronchial stump
Ruffini et al. (2001) Eur J Cardiothoracic Surg, Italy [15] Retrospective case-control study (level III)	n = 1221 Lung resection (all types) for lung cancer 1993–1999 Fluid regimen: Crystalloid Intraoperative fluid restriction of <1.5 l 1.5 l of crystalloid over 48 h postoperatively	PALI/ARDS: 27 (2.2%) (Using consensus criteria [4])	No evidence of an association between fluid intake and PALI/ARDS	Primary data relating to fluid administration is unpublished Intraoperative fluid restriction of <1500 ml may account for the low incidence (2.2%) of PALI/ARDS
Kutlu et al. (2000). Ann Thorac Surg, UK [14] Retrospective case-control study (level III)	n = 1139 Lung resections (all types) 1991–1997 Fluid regimen: Not defined	PALI/ARDS: 45 (3.9%) (Using consensus criteria [4])	No evidence of fluids as a risk factor	Primary data relating to fluid administration is unpublished
Van der Werff et al. (1997) Chest, Netherlands [12] Retrospective case-control study (level III)	n = 197 Pneumonectomy 1989–1995 Fluid regimen: Not defined	Postoperative pulmonary oedema (PPE): 29 (14.7%)	No evidence of fluids as a risk factor for PPE	Primary data relating to fluid administration is unpublished Consensus criteria [4] are not used therefore incidence of PALI/ARDS may be over-/under-estimated
Hayes et al. (1995) Thorax, UK [13] Retrospective case-control study (level III)	n = 469 Lung resections (all types) 1991–1994 Fluid regimen: Not defined	PALI/ARDS: 22 (5.1%) (Using consensus criteria [4])	No evidence of fluids as a risk factor	Primary data relating to fluid administration is unpublished
Zeldin et al. (1984) J Thorac Cardiovasc Surg, USA [17] Retrospective unmatched case-control study of selected patients (level IV)	n = 25 Pneumonectomy	Pulmonary oedema: 10 Uncomplicated pneumonectomy: 15	Fluid intake (pulmonary oedema group): 4913 ± 1169 ml (mean ± SD) Fluid intake (uncomplicated pneumonectomy group): 3483 ± 984 ml	Consensus criteria [4] post-date the study Severity of pulmonary oedema is not described Cases and controls are not matched
Zeldin et al. (1984) J Thorac Cardiovasc Surg, USA [18] Experimental canine animal study	n = 13 A (n = 8): 100 ml/kg crystalloid immediately prior to right pneumonectomy followed by >100 ml/kg postoperative fluid balance B (n = 5): 50 ml/kg immediately prior to right pneumonectomy followed by 50 ml/kg intraoperatively followed by <100 ml/kg postoperative fluid balance C (n = 4): 100 ml/kg immediately prior to sham thoracotomy followed by >100 ml/kg postoperative fluid balance	Pulmonary oedema: 6 2 dogs in group A died prematurely (Left lung weight/right lung weight ratio >1) (Left lung/body weight ratio >0.68) All dogs received 4 ml of fluid for every millilitre of intraoperative blood loss	Group A: all surviving dogs developed pulmonary oedema Group B: no dogs developed pulmonary oedema Group C: no dogs developed pulmonary oedema	Fluid regimens described are not directly comparable to medical practice Severity of pulmonary oedema is not described in clinically-translatable terms Powerful demonstration of the pathophysiology of pneumonectomy exposing the remaining lung to a risk of pulmonary oedema Evidence of a link between post-pneumonectomy pulmonary oedema and large fluid load in mammals

UVA: univariate analysis; MVA: multivariate analysis; PALI: postoperative acute lung injury; ARDS: acute respiratory distress syndrome; NSCLC: non-small cell lung cancer.

RESULTS

The best evidence available to answer this question are the studies by Licker et al. [2, 3]. Both used consensus guideline criteria [4] for the diagnosis of PALI/ARDS and demonstrated an association between increasing the volume of fluid administration in the intraoperative and postoperative periods and the incidence of PALI/ARDS on multivariate analysis (MVA) with odds ratios (ORs) of 1.42 (95% CI 1.09–4.32, P = 0.011) [2] and 2.91 (1.9–7.4, P = 0.001) [3]. In non-PALI/ARDS cases, the mean intraoperative fluid infusion volume was significantly less (1.22 l (1.17–1.26) vs 1.68 l (1.46–1.9) P = 0.005), fluid balance over the first 24 postoperative hours was significantly less (1.52 l positive (1.44–1.60) vs 2.0 l positive (1.6–2.4) P = 0.026) and cumulated intra- and postoperative fluid infusion was also significantly less (2.0 ml/kg/h (1.7–2.3) vs 2.6 ml/kg/h (2.3–2.9) P = 0.003) [3]. Total fluid infusion in the first 24 h postoperatively was not significantly different (PALI/ARDS group 2.1 l (1.85–2.33), non-PALI/ARDS group 1.85 l (1.79–1.91) P = 0.075) [3]. This data set provides evidence that the difference between fluid therapy associated with an increased incidence of PALI/ARDS (i.e. ‘liberal’) and fluid therapy associated with no increased incidence (i.e. ‘conservative’) is narrow, although it is not possible to provide an exact definition of liberal fluid therapy from these data. Alam et al. [5] also demonstrated a significant association on MVA between increasing fluid administration and the outcome ‘primary lung injury’ (OR 1.2 (1–1.4), P = 0.02) for every 500 ml increase in perioperative fluid administration; however, this study was limited by the use of non-standardized diagnostic criteria for PALI/ARDS, raising the potential for an over- or under-diagnosis of the condition.

Several of the other studies analysed were limited in that fluids were not directly assessed against PALI/ARDS; composite outcome measures such as ‘all respiratory’ or 'all major’ complications were used [6–10]. Others were limited for the purposes of comparison by using non-standardized outcomes related to PALI/ARDS, e.g. ‘primary lung injury’ and ‘postoperative pulmonary oedema’ [5, 9–12] rather than consensus criteria [4]. Where no association is reported, several papers have not published the amount of fluid administered [10, 12–15]. Blank et al. [6] may have shown an association between fluid intake and the outcome ‘all respiratory complications’ were the cohort of PALI/ARDS cases larger (n = 9) [16].

The published data show that liberal fluid therapy is associated with a higher incidence of PALI/ARDS; however, a causal relationship cannot be inferred; PALI/ARDS is clearly a multifactorial disease process with liberal fluid therapy being one of several associated risk factors (e.g. blood transfusion, hypotensive haemorrhage, prolonged duration of one-lung ventilation and ventilator-induced lung injury). It is thought that the development of PALI/ARDS may require two or more of these ‘hits’ [17]. The evidence presented here follows on from the 1984 unmatched case–control study [18] of 25 patients which suggested that a liberal vs judicious perioperative fluid load is associated with pulmonary oedema following pneumonectomy (4913 ± 1169 ml (n = 10 pulmonary oedema cases) vs 3483 ± 984 ml (n = 15 controls) (mean ± SD)). This study also described a canine experimental model of right pneumonectomy with a 48 h period of postoperative monitoring that assessed a liberal crystalloid perioperative fluid load (100 ml/kg rapidly infused immediately preoperatively followed by a >100 ml/kg postoperative fluid balance (n = 8)) vs a judicious perioperative fluid load (50 ml/kg rapidly infused immediately preoperatively followed by 50 ml/kg rapidly infused intraoperatively followed by a <100 ml/kg postoperative fluid balance [n = 5]) vs a control group with a 100 ml/kg preoperative rapid fluid load followed by sham thoracotomy and >100 ml/kg postoperative fluid balance. All dogs in the liberal fluid load group developed pulmonary oedema, whereas no dogs in the judicious fluid load or control group developed the condition. The control group data powerfully demonstrate that the perioperative pathophysiology of pneumonectomy exposes the remaining lung to a risk of pulmonary oedema in mammals.

CLINICAL BOTTOM LINE

PALI/ARDS has a mortality of >50% [19]; therefore, strategies to reduce its incidence are of great interest. On this best evidence presented, we recommend a conservative strategy of administration of maintenance fluids at 1–2 ml/kg/h in the intra- and post-operative periods and that a positive fluid balance of 1.5 l should not be exceeded, to mitigate the risk of multifactorial PALI/ARDS. Caution should be exercised with regard to silent hypovolaemia, impaired oxygen delivery and acute kidney injury [20]. If the fluid balance exceeds this threshold, a high index of suspicion for PALI/ARDS should be adopted and an escalation of the level of care should be considered. If a patient develops signs of hypoperfusion after these thresholds are exceeded, inotropic/vasopressor support should be considered.

Conflict of interest: none declared.