Fluid bolus resuscitation is integral to developed world

Transcription

1 Fluid Overload at 48 Hours Is Associated With Respiratory Morbidity but Not Mortality in a General PICU: Retrospective Cohort Study* Lynn Sinitsky, MB, MRCPCH 1 ; David Walls, BSc 2 ; Simon Nadel, MB, FRCP 1 ; David P. Inwald, FRCPCH, PhD 2 Objective: Recent evidence suggests that fluid overload may be deleterious to critically ill children. The purpose of this study was to investigate the association of early fluid overload with respiratory morbidity and mortality in patients admitted to a general PICU. Design: Retrospective cohort study. Setting: Single, tertiary referral PICU. Patients: Six hundred thirty-six patients aged 0 16 years invasively ventilated at 48 hours post admission, admitted between April 1, 2009, and March 31, Measurements and Main Results: Data collected included demographics, diagnosis, Pediatric Index of Mortality 2 score, and fluid overload percent at 48 hours from admission. Fluid overload percent was calculated as (cumulative fluid in cumulative fluid out (L))/ hospital admission weight (kg) 100%. Outcome measures were oxygenation index at 48 hours from admission, death, and invasive ventilation days in survivors. Data are reported as median (interquartile range) and were analyzed using nonparametric tests. The median age was 1.05 years ( yr). Fifty-three patients (8%) died. Median duration of ventilation in survivors was 5 days (3 8 d). Fluid overload percent correlated significantly with oxygenation index (Spearman ρ, 0.318; p < ) and with invasive ventilation days in survivors (Spearman ρ, 0.274; p < ). There was no significant difference in fluid overload percent between survivors and nonsurvivors. Regression analysis demonstrated that fluid overload *See also p Paediatric Intensive Care Unit, Imperial College Healthcare NHS Trust, St Mary s Hospital, London, United Kingdom. 2 Patient Care & Clinical Informatics, Healthcare, Philips, Guildford, United Kingdom. This study was performed at Paediatric Intensive Care Unit, Imperial College Healthcare NHS Trust, St Mary s Hospital, London, United Kingdom. Mr. Nadel received support from the National Institute for Health Research Biomedical Research Centre. Dr. Inwald received support for article research from the Imperial College Biomedical Research Centre. Dr. Nadel consulted and lectured for Novartis Vaccines. The remaining authors have disclosed that they do not have any potential conflicts of interest. For information regarding this article, d.inwald@imperial.ac.uk Copyright 2015 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies DOI: /PCC percent was a significant predictor of both oxygenation index at 48 hours (p < 0.001) and invasive ventilation days (p = 0.002). Conclusions: Fluid overload at 48 hours was associated with oxygenation index at 48 hours and invasive ventilation days in survivors in a general PICU population. There was no association of fluid overload at 48 hours with mortality. (Pediatr Crit Care Med 2015; 16: ) Key Words: fluid; fluid overload; mechanical ventilation; pediatric; respiratory failure Fluid bolus resuscitation is integral to developed world management of children presenting with signs of severe sepsis, despite the lack of high-grade evidence to support its use (1). Enthusiastic fluid resuscitation, along with ongoing maintenance fluid administration, commonly leads to positive fluid balance, sometimes with fluid overload, in children admitted to pediatric intensive care. Increasing evidence is accumulating that fluid overload in critical illness is deleterious in both children and adults (2 6). In children with acute lung injury (ALI), early fluid overload has been independently associated with mortality and with duration of mechanical ventilation (7, 8). Analysis of the 109 patients in the pediatric arm of the Calfactant in Acute Respiratory Distress Syndrome trial (9), a multicenter randomized placebocontrolled trial of endotracheally instilled calfactant versus placebo in patients intubated for ALI, also demonstrated that fluid overload was associated with both mortality and respiratory morbidity. In adults, the Fluid and Catheter Treatment Trial (10), a large multicentered randomized control study, comparing liberal and conservative fluid management strategies, demonstrated that patients with ALI managed with a conservative fluid strategy had improved oxygenation, decreased duration of mechanical ventilation, and reduced ICU stay. If there is an effect of fluid overload on pulmonary dysfunction, fluid overload would be expected to be detrimental in all mechanically ventilated patients, whether or not they have ALI. Accordingly, a recent small study of 80 patients has also shown an association between fluid overload, impaired oxygenation, and increased duration of ventilation in a general PICU population (11). Pediatric Critical Care Medicine 205

2 Sinitsky et al In contrast, the post hoc analysis of the Pediatric Acute Lung Injury and Sepsis Investigators (PALISI) multicenter weaning study did not find any association between cumulative fluid balance, duration of weaning, and extubation outcome (12). Despite this, subgroup analysis demonstrated that the group with the highest cumulative fluid balance had an increased duration of weaning. These inconsistent findings may in part reflect the difficulties of post hoc analysis in a study not designed to take into account fluid overload on the outcomes of interest. Thus, most studies to date suggest that fluid overload in critically ill children is detrimental. However, the current evidence base is fraught with difficulties, including limited investigation of specific patient groups, small cohorts, inaccurate data collection, and post hoc analyses. Because of these uncertainties, we decided to investigate the association of early fluid overload with oxygenation index (OI), length of ventilation, and mortality in mechanically ventilated children in a large general PICU population. MATERIALS AND METHODS It is a single-center retrospective anonymized database study in patients admitted to the PICU at St Mary s Hospital, London, United Kingdom, over a 4-year period, April 1, 2009, to March 31, St Mary s PICU is a general medical and surgical ICU admitting children with a wide variety of pathology. There is no cardiac surgical program. Seventy percent of admissions to the PICU are transfers from secondary care centers (District General Hospitals) located in London and the South-East of England. Local research ethics approval for the study was obtained from Imperial College Healthcare National Health Service Trust Joint Research Compliance Office (Study 13SM1803), and the need for informed consent was waived. All patients who are 0 16 years old and were invasively ventilated for at least 24 hours and remained on invasive ventilation at 48 hours post admission were eligible for inclusion into the study. There were no exclusion criteria. The PICU clinical information system database (Intellivue Clinical Information Portfolio [ICIP], Philips, United Kingdom) and the Paediatric Intensive Care Audit Network database (PICANet, Leeds, United Kingdom) were interrogated to yield prospectively collected data. ICIP is the electronic patient record, which includes clinical records, physiological charting including fluid balance, and laboratory data. PICANet is the national U.K. PICU audit, on which quality-assured data are recorded, including demographics, diagnosis, PIM2 scoring, PICU interventions, and outcome. Data collection included demographic characteristics, diagnosis, hospital admission weight, PIM2 score, ventilator variables (mode of ventilation, mean airway pressure, and Fio 2 ), PICU interventions (vasoactive drugs and continuous renal replacement therapy [CRRT]), duration of invasive ventilation, daily fluid intake and output from admission, and arterial blood gas measurements. Physiological and diagnostic data including fluid intake and output are entered manually on an hourly basis by the bedside nurse; laboratory, blood gas, and ventilator data are entered automatically via electronic interface. Daily fluid intake included maintenance IV fluids, parenteral and enteral nutrition, fluid boluses, blood and blood products, and medication. Daily maintenance fluids were calculated according to the Holliday-Segar formula (13). Daily fluid output included urine output, gastric aspirate, stool, and losses from other body cavities. Insensible losses were not calculated. OI was calculated only if blood gas analysis had been performed on blood taken from an indwelling arterial catheter. OI was calculated at 48 hours post admission as follows: Fio 2 mean airway pressure 100/Pao 2. Fluid overload percent (%FO) at 48 hours post admission was calculated as follows: cumulative fluid in from admission to 48 hours (L) cumulative fluid out from admission to 48 hours (L)/hospital admission weight (kg) 100%. Statistical analysis was performed to investigate the relationship between %FO at 48 hours and three outcome measures: 1) OI at 48 hours from admission; 2) invasive ventilation days in survivors; and 3) death. The Spearman rank correlation coefficient (Spearman ρ) was used to assess the relationships between variables. The Mann-Whitney U test (MWU) was used to test whether differences existed in variables between survivors and nonsurvivors. The Kruskal-Wallis test was used to test whether differences existed in variables in analyses with more than two groups. The chi-square test was used to test for differences in proportions between groups. Linear regression analysis was used to explore the relationship between %FO and OI at 48 hours from admission and %FO and invasive ventilation days in survivors. Predictor variables selected to be included in the models were those found to be significantly correlated with or associated with the dependent variable in the initial analysis (p < 0.1). PICU interventions (vasoactive drugs, CRRT) were not included in these models due to collinearity with PIM2 score. Some previous studies have used bandings of %FO to look at the relationship between %FO and various outcome measures, noting that a positive fluid balance more than 15% increased the risk for increased morbidity and pulmonary dysfunction in a general PICU population (11). We therefore also divided %FO into bands (0% to < 5%, 5% to < 10%, 10% to < 15%, and 15%) post hoc for parts of our analysis. Data were nonparametric and are presented as median and interquartile range. SPSS Statistics version 20.0 (IBM, Armonk, NY) was used for the analysis. RESULTS Over the 4-year study period, 1,293 children were admitted to our PICU, and 636 children (49%) met eligibility criteria and were included in the study with a variety of diagnoses (Table 1). OI data were available for 453 eligible patients (71%). The remainder did not have an arterial catheter in situ at 48 hours. Three hundred sixty-one patients (57%) were male, and median age was 1.05 year ( yr). Three hundred seventy-three patients (59%) had a primary respiratory diagnosis (Table 1). Ninety patients (14%) were receiving high-frequency oscillation ventilation; the remainder were conventionally ventilated at 48 hours. Median PIM2 score was 6% (2 12) March 2015 Volume 16 Number 3

3 Table 1. Patient Characteristics: Diagnostic Category at Admission (n = 636) Diagnostic Category n (% of Total) Mortality (%) Respiratory 373 (59) 28 (8) Sepsis 119 (19) 9 (8) Neurologic 59 (9) 3 (5) Elective postsurgical 23 (4) 1 (4) Cardiac (including 19 (3) 8 (42) post cardiac arrest) Other 22 (3) 3 (14) Trauma 12 (2) 0 (0) Metabolic 9 (1) 1 (11) Overall, 53 patients (8%) died. In survivors, the median duration of invasive ventilation was 5 days (3 8 d). Thirty-four patients (5%) received CRRT and 320 (50%) received vasoactive drugs. At 48 hours from admission, the median %FO was 7.2% ( %) and median OI was 4.8 ( ). Data on demographics, severity of illness, PICU interventions, and outcomes across the fluid overload bandings are shown in Table 2. This demonstrates that %FO was significantly associated with younger age, higher PIM2 score, higher requirement for vasoactive drugs and CRRT, higher OI at 48 hours, and days of invasive ventilation in survivors. There was no association with %FO and mortality. Looking at the population as a whole, there was a significant correlation between %FO and OI at 48 hours (Spearman ρ, 0.318; p < 0.001). There was no significant correlation or association between OI at 48 hours and PIM2, age, or gender (results not shown). Diagnostic category was, however, associated with OI at 48 hours (Kruskal-Wallis test, p < 0.001), with median OI varying from 2 ( ) in the sepsis group to 6.5 ( ) in the respiratory group. Diagnostic category was not associated with %FO (p = ns, Kruskal Wallis, results not shown). Linear regression was therefore performed including %FO and diagnostic category as predictor variables in a model with OI at 48 hours as the dependent variable. This indicated that the overall model was a fair fit for the data (F (2,450) = 31.19; p < 0.001; R 2 = 0.122), with both %FO (p < 0.001; β = 0.26; se = 0.05) and diagnostic group (p < 0.001; β = 0.94; se = 0.16) significant predictors. Looking at invasive ventilation days in survivors, there was a significant relationship between %FO bandings and invasive ventilation days (Kruskal-Wallis test, p < ) (Fig. 1 and Table 2). Invasive ventilation days in survivors also correlated significantly with PIM2 (Spearman ρ, 0.148; p < ) and %FO (Spearman ρ, 0.274; p < ), but not with age. There was also an association between invasive ventilation days and diagnostic group (MWU, p = 0.032) but not gender. A linear regression model was built, using invasive ventilation days as the dependent variable. Predictor variables were PIM2, %FO, and diagnostic group. The regression analysis demonstrated that the overall model was a fair fit for the data (F (3,578) = 14.91; p < 0.001; R 2 = 0.07), with %FO a significant predictor of invasive ventilation days (p = 0.002; β = 0.14; se = 0.05), along with PIM2 (p < 0.001; β = 17.1; se = 3.3) and diagnostic group (p = 0.004; β = 0.46; se = 0.16). Differences between survivors and nonsurvivors are shown in Table 3. Although there were significant differences in age, sex, PIM2, and PICU interventions between survivors and nonsurvivors, there was no significant difference in %FO (MWU, p = 0.23) (Table 3). Looking across a range of different bandings of %FO, there was a tendency for increased mortality in the higher banding, but this was not statistically significant (chi-square test, p = 0.53) (Table 2). Diagnostic group was significantly associated with mortality (chi-square test, p < 0.001), ranging from 42% in the cardiac/post cardiac arrest group to 0% in the trauma group (Table 1). DISCUSSION This is the largest retrospective study to date investigating fluid overload and outcome in a cohort of undifferentiated critically Table 2. Comparison of All Patients by Fluid Overload Percent Bandings (n = 636) Characteristic Fluid Overload, 0% to < 5% (n = 194) Fluid Overload, 5% to < 10% (n = 234) Fluid Overload, 10% to < 15% (n = 101) Fluid Overload, 15% (n = 107) Statistical Test Age (decimal years) 1.92 ( ) 1.24 ( ) 1.1 ( ) 0.93 ( ) p < (KW) Male sex 110 (57%) 139 (59%) 60 (60%) 52 (49%) p = 0.28 (χ 2 test) PIM ( ) 0.06 ( ) 0.08 ( ) 0.09 ( ) p < (KW) Vasoactive drugs 74 (38%) 99 (42%) 74 (73%) 73 (68%) p < (χ 2 test) Continuous renal 7 (4%) 9 (4%) 5 (5%) 13 (13%) p = (χ 2 test) replacement therapy Oxygenation index at 0.06 ( ) 4.2 ( ) 6.7 ( ) 7.1 ( ) p < (KW) 48 hr Mortality 13 (6.7%) 18 (7.6%) 11 (10.9%) 11 (11.5%) p = 0.53 (χ 2 test) Invasive ventilation days 5 (3 7) 5 (4 8) 6 (5 12) 8 (5 13) p < (KW) KW = Kruskal-Wallis test. Invasive ventilation days are shown for survivors only (n = 583). p values are reported from statistical tests comparing data across each row. Pediatric Critical Care Medicine 207

4 Sinitsky et al Figure 1. Duration of ventilation in survivors according to different levels of fluid overload at 48 hr from admission. Medians and interquartile range are shown. The relationship between fluid overload percent and invasive ventilation days in survivors was significant, despite the wide interquartile range (Kruskal-Wallis test, p < 0.001). ill children. Our study demonstrated that although degree of fluid overload at 48 hours is associated with both OI at 48 hours and length of ventilation in survivors, there was no association with mortality. Despite weak R 2 values, the association between fluid overload and OI and length of invasive ventilation was significant and confirmed in the regression models. Our study is consistent with Arikan et al s (11) smaller retrospective study of fluid overload in 80 general PICU patients. The patients included in the study by Arikan et al (11) and our study were both representative of admissions to a general PICU, with a similar spread of diagnoses. Our cohort was younger, had a lower OI, and was ventilated for a shorter period of time. Primary respiratory pathology was the most common diagnosis in both studies, and in both, there was a significant association between fluid overload and pulmonary dysfunction. The study by Arikan et al (11) investigated fluid overload up to 14 days from admission. We chose to focus on early fluid overload as most critically ill children in the United Kingdom spend less than 7 days in PICU (14). Also, accurate measurement of fluid balance is an ongoing nursing challenge (15, 16), particularly as insensible losses cannot be measured and as long-stay patients in particular often do not have urinary catheters. It is possible that in some other studies of fluid overload in ventilated children, in which many patients reached their peak fluid overload after day 5 of admission, this is due to inaccurate measurement. For example, in the PALISI study (12), children were found to be a mean of 136 ml/kg positive at the time of extubation, with a range of 358 to +2,228 ml/ kg. Such a wide range casts doubt on the validity of the overall findings. Our own data demonstrated that fluid overload progressively increased over admission in long-stay patients (data not shown). We suspect that this is due to inaccurate charting of fluid balance once patients no longer have a urinary catheter and stools become solid, rather than ongoing excess fluid administration beyond the first 48 hours. It is for this reason that we decided to focus our study on mechanically ventilated patients in the first 48 hours of admission and in patients with an arterial catheter in situ in our unit, such patients always have a urinary catheter in situ and fluid balance calculation is therefore at its most reliable. The %FO at 48 hours in our study is similar to that found in other studies which have also chosen to focus on early rather than late fluid overload; for example, the 75% quartile of fluid overload was 11% on day 2 in the study by Arikan et al (11) and the mean cumulative fluid overload at day 3 in the study by Valentine et al (8) was 8.5% ± 10.5%. The observation that %FO is associated with vasoactive drug use is interesting, and the significance is not clear. There are a two possible explanations first, that fluid bolus therapy is being given at the point the Starling curve is flat, or in other words, a failure to respond to fluids alerts the clinician that vasoactive drugs are required; or second, that patients on vasoactive drugs are more unwell with more capillary leak and simply require increased fluids. Without data on timing of fluid versus vasoactive drug therapy, it is not possible to comment further. Future prospective studies with improved hemodynamic monitoring may help to dissect this association (17). A strength of this study is its relevance and general applicability to critically ill children. Our large cohort included patients with a broad range of diagnoses, other than postoperative cardiac patients, as our unit does not have a cardiac surgical or extracorporeal membrane oxygenation program. The study also demonstrates proof of concept for utilization of a PICU clinical information system for retrospective studies of this nature. Table 3. Comparison of Survivors With Nonsurvivors (n = 636) Characteristic Survivors (n = 583) Nonsurvivors (n = 53) Statistical Test Age (decimal years) 0.98 ( ) 4.22 ( ) p < (MWU) Male sex 338/583 (58%) 23/53 (43%) p = 0.04 (χ 2 test) PIM ( ) 0.12 ( ) p < (MWU) Vasoactive drugs 276/583 (47%) 44/53 (83%) p < (χ 2 test) Continuous renal replacement therapy 24/583 (4%) 10/53 (19%) p < (χ 2 test) Fluid overload percent at 48 hr 7.2 ( ) 7.7 ( ) p = 0.23 (MWU) MWU = Mann-Whitney U test. Continuous variables are shown as median (interquartile range). p values are reported from appropriate statistical tests comparing data across each row March 2015 Volume 16 Number 3

5 Limitations The main limitation of our study was the failure to include fluid input and output prior to PICU admission due to practical constraints. It is most likely that this resulted in an underestimate of fluid overload, as many patients would likely have received large volumes of fluid in excess of output during resuscitation and/or stabilization prior to PICU admission (18). Fluid overload may also been underestimated in patients with cardiac failure or acute kidney injury as no account was taken of disease-related fluid accumulation prior to admission. The use of a fluid balance based rather than a weight-based calculation of fluid overload might be another limitation in our study. The weight used for our fluid overload calculation was hospital admission weight. It is not our practice to reweigh patients in the PICU. Such practice would have enabled a weight-based determination of fluid overload for comparison. However, although %FO has previously been shown to differ when it is calculated by fluid balance or weight, fluid balance based and weight-based methods correlate very highly with each other and share similar ability to predict death in patients requiring CRRT (19). Furthermore, as the data demonstrate, there are many factors at play in the complex relationship between %FO, severity of illness, PICU interventions, and outcome. It is hardly surprising that PIM2 score is associated with both %FO in survivors and with mortality, for example. The association between age and mortality is less expected but may have been confounded by the higher prevalence of comorbidities in older children, rather than because of a direct effect of age. Comorbidities, including neurological and respiratory problems, are much more common in older children invasively ventilated in PICU and are associated with mortality. In the group described, diagnostic data indicated that of the 53 children who died, 39 had major comorbidities including 22 with severe neurological impairment. The failure to demonstrate an association between fluid overload and mortality may also have been due to the study being underpowered to detect such an association. No formal power calculation was performed; it is possible that a larger cohort would demonstrate an association. The inverse association of age with %FO in survivors, with younger patients having a tendency to be more fluid overloaded, is also interesting and may be either due to a tendency for the Holliday-Segar formula to overestimate maintenance fluid requirements in ventilated infants or due to increased fluid bolus therapy in younger patients. A prospective study separating out maintenance from fluid bolus therapy would be necessary to determine which of these explanations are the reasons for this observation. CONCLUSIONS Fluid overload at 48 hours is associated with OI at 48 hours and length of ventilation in a general PICU population. Overall, our findings support the need for a prospective study of fluid overload in PICU, including fluid bolus resuscitation given prior to PICU admission, to elucidate the relationship between fluid overload and outcome. ACKNOWLEDGEMENT We thank Dr. Alex Couto Alves for statistical advice. REFERENCES 1. Brierley J, Carcillo JA, Choong K, et al: Clinical practice parameters for hemodynamic support of pediatric and neonatal septic shock: 2007 update from the American College of Critical Care Medicine. Crit Care Med 2009; 37: Foland JA, Fortenberry JD, Warshaw BL, et al: Fluid overload before continuous hemofiltration and survival in critically ill children: A retrospective analysis. Crit Care Med 2004; 32: Goldstein SL, Somers MJ, Baum MA, et al: Pediatric patients with multi-organ dysfunction syndrome receiving continuous renal replacement therapy. Kidney Int 2005; 67: Sutherland SM, Zappitelli M, Alexander SR, et al: Fluid overload and mortality in children receiving continuous renal replacement therapy: The prospective pediatric continuous renal replacement therapy registry. Am J Kidney Dis 2010; 55: Michael M, Kuehnle I, Goldstein SL: Fluid overload and acute renal failure in pediatric stem cell transplant patients. Pediatr Nephrol 2004; 19: Hazle MA, Gajarski RJ, Yu S, et al: Fluid overload in infants following congenital heart surgery. Pediatr Crit Care Med 2013; 14: Flori HR, Church G, Liu KD, et al: Positive fluid balance is associated with higher mortality and prolonged mechanical ventilation in pediatric patients with acute lung injury. Crit Care Res Pract 2011; 2011: Valentine SL, Sapru A, Higgerson RA, et al; Pediatric Acute Lung Injury and Sepsis Investigator s (PALISI) Network; Acute Respiratory Distress Syndrome Clinical Research Network (ARDSNet): Fluid balance in critically ill children with acute lung injury. Crit Care Med 2012; 40: Willson DF, Thomas NJ, Tamburro R, et al; Pediatric Acute Lung and Sepsis Investigators Network: The relationship of fluid administration to outcome in the pediatric calfactant in acute respiratory distress syndrome trial. Pediatr Crit Care Med 2013; 14: National Heart, Lung, and Blood Institute Acute Respiratory Distress Syndrome (ARDS) Clinical Trials Network; Wiedemann HP, Wheeler AP, Bernard GR, et al: Comparison of two fluidmanagement strategies in acute lung injury. N Engl J Med 2006; 354: Arikan AA, Zappitelli M, Goldstein SL, et al: Fluid overload is associated with impaired oxygenation and morbidity in critically ill children. Pediatr Crit Care Med 2012; 13: Randolph AG, Forbes PW, Gedeit RG, et al; Pediatric Acute Lung Injury & Sepsis Investigators (PALISI) Network: Cumulative fluid intake minus output is not associated with ventilator weaning duration or extubation outcomes in children. Pediatr Crit Care Med 2005; 6: Holliday MA, Segar WE: The maintenance need for water in parenteral fluid therapy. Pediatrics 1957; 19: Paediatric Intensive Care Audit Network National Report UK, Universities of Leeds and Leicester, Ware LB: Pathophysiology of acute lung injury and the acute respiratory distress syndrome. Semin Respir Crit Care Med 2006; 27: van Asperen Y, Brand PL, Bekhof J: Reliability of the fluid balance in neonates. Acta Paediatr 2012; 101: McLean JR, Inwald DP: The utility of stroke volume variability as a predictor of fluid responsiveness in critically ill children: A pilot study. Intensive Care Med 2014; 40: Lampariello S, Clement M, Aralihond AP, et al: Stabilisation of critically ill children at the district general hospital prior to intensive care retrieval: A snapshot of current practice. Arch Dis Child 2010; 95: Selewski DT, Cornell TT, Lombel RM, et al: Weight-based determination of fluid overload status and mortality in pediatric intensive care unit patients requiring continuous renal replacement therapy. Intensive Care Med 2011; 37: Pediatric Critical Care Medicine 209