Near-Normal Values of Extravascular Lung Water in Children

Objectives: To define near-normal values of extravascular lung water indexed to body weight in children. Design: Prospective multicenter observational study. Setting: Medical/surgical PICUs of 5 multinational hospitals. Patients: Fifty-eight children with a median age of 4 years (range 1 month to 17 year) with heterogeneous PICU admission diagnoses were included. Extravascular lung water measurements from these children were collected after resolution of their illness. Obtained values were indexed to actual body weight and height and subsequently related to age. Interventions: None. Measurements and Main Results: Extravascular lung water indexed to body weight correlated with age (r2 = 0.7) and could be categorized in three-age groups consisting of significantly different median extravascular lung water indexed to body weight values (5th–95th percentile): less than 1 year, 9–29 mL/kg; 1–5 years, 7–25 mL/kg; and 5–17 years, 5–13 mL/kg. Extravascular lung water indexed to height did not correlate to age and resulted in an age-independent near-normal value of less than 315 mL/m. Conclusions: Younger children have higher values of extravascular lung water indexed to actual body weight. Age categorized near-normal values of extravascular lung water indexed to body weight are presented for possible clinical use. Furthermore, we suggest to index extravascular lung water to height, which seems to be age independent.

E xtravascular lung water (EVLW) can be measured at the bedside using the transpulmonary thermodilution method incorporated in the PiCCO device (Pulsion, Munich, Germany). EVLW reflects the amount of pulmonary edema with acceptable accuracy. It is associated with an increased risk of mortality in both adults and children and may be used to guide fluid therapy (1)(2)(3)(4)(5)(6)(7)(8)(9).
Traditionally, EVLW is indexed using body weight (EVLWI [mL/kg]). For the adult population, values below 10 mL/kg are considered normal (8,10,11). However, higher values of EVLWI have been described in young children (4,11,12). Since EVLW is related to lung mass, the reason for these high values of EVLWI is in part explained by the changing relation between lung mass and body weight during human growth and development, specifically in the first years of life (13). In the young, the ratio of lung mass (and thus the amount of lung water) to body weight is higher (13). Up to now, there are still no age-related normal values available, rendering the interpretation and clinical use of this variable impractical in pediatric patients.
Adult studies have shown that indexing EVLW to predicted body weight produces a more physiologic and useful value (14,15). Others suggest that indexing EVLW to height is superior to the aforementioned weight-based methods (16). We hypothesized that 1) EVLW is higher in younger children and 2) weightindexed EVLW is age dependent, whereas height-indexed EVLW might be age independent and thus more easy to use in clinical practice. This study was designed to collect EVLW values from children, admitted to a PICU after resolution of their illness. The obtained values were used to define age-related near-normal values of indexed lung water in children.
www.pccmjournal.org e29 no arterial PiCCO catheters or central venous catheter was inserted because of the study. Therefore, the medical ethical committee of the Radboud University Medical Center waived the need for informed consent in concordance with Dutch legislation. Data were collected in PICUs in Tunis (Tunisia), Rome (Italy), Santa Barbara (CA), Hamburg (Germany), and Nijmegen (The Netherlands).
We included children (0-17 yr) admitted to a PICU and connected to a PiCCO monitor for routine hemodynamic monitoring. EVLW values were eligible for the study if they were measured under stable hemodynamic condition and after resolution of the acute illness. Resolution of respiratory derangements was reflected by 1) spontaneous breathing or mechanical ventilation with minimal support (pressure support maximum 10 cm H 2 O or likewise, positive end-expiratory pressure ≤ 5 cm H 2 O, and Fio 2 ≤ 40%); 2) noninvasive oxygen saturation (Sao 2 ) more than 92%; 3) no clinical signs of pulmonary edema; and 4) no signs of increased work of breathing. Stable hemodynamics were reflected by 1) normal sinus rhythm and normal blood pressure; 2) normal renal function; 3) no clinical signs of edema; and 4) either no or minimal vasoactive support (dopamine or dobutamine ≤ 5 μg/kg/min, epinephrine or norepinephrine < 0.1 μg/kg/min, and milrinone < 0.25 μg/kg/min or enoximone < 10 μg/kg/min).
Exclusion criteria were clinical signs of (pulmonary) edema or increased work of breathing and an abnormal chest radiograph. Additionally, measurements in the presence of an intracardiac or extracardiac left-to-right or right-to-left shunt were excluded.
The PiCCO system requires a central venous catheter and a special thermistor-equipped arterial catheter inserted in the femoral artery. The three main recorded variables are cardiac output (CO), global end-diastolic blood volume (GEDV), which reflects preload, and EVLW. The technical background of the PiCCO system and the calculation of the various hemodynamic variables have been described in detail elsewhere (4,17). The mean value of multiple (at least three) measurements was recorded. Individual measurements with technical defects reflected by extreme abnormal values or an abnormal thermodilution curve were discarded. Measurements were performed using various PiCCO devices (PiCCOplus, PiCCO 2 , or PiCCO module for use with Hewlett Packard/Philips Monitoring System; Philips Healthcare, Best, The Netherlands), and only the absolute values were recorded. Other recorded data included, among others, heart rate, blood pressure, central venous pressure, respiratory rate, oxygen saturation, and blood gas results.
Body surface area (BSA) was calculated by the Haycock method using actual weight (18). CO  All data were tested for normality. In case of normal distribution, a t test (or when appropriate analysis of variance) was used; otherwise, Mann-Whitney U test was performed. A p value of less than 0.05 was considered significant. Normal values were constructed using the 5th-95th percentile values. Statistical calculations were made using Medcalc (Mariakerke, Belgium).

RESULTS
Fifty-eight children were included in this study of which 11 children were younger than 1 year ( Table 1). Of the children up to 1 year, the median age was 5 months, and three children were younger than 3 months. Forty-two children were breathing spontaneously while 16 were being mechanically ventilated. Two  Figure 1A shows the nonlinear relation between EVLW and body weight, whereas Figure 1B shows the relation between EVLW and body height. Figure 2 shows the relation between age and EVLW indexed to actual body weight. Open circles represent ventilated children, whereas closed circles represent spontaneously breathing children. Figures 3 and 4 reflect EVLW indexed to either actual body weight or height divided per age group. There is a significant difference in EVLW indexed to body weight: the older age groups have significant lower values of EVLWI compared with the younger age groups. However, there is no difference in EVLW indexed to height between the different age groups.
There were no significant differences between mechanically ventilated or spontaneously breathing children with regard to A-a gradient, Pao 2 /Fio 2 ratio, EVLWI, CI, or GEDVI. There were also no differences between boys and girls with respect to age, weight, height, CI, GEDVI, or EVLWI. We did not observe   www.pccmjournal.org e31 any influence of the site of the central venous catheter (internal jugular, subclavian, or femoral vein) on the results of the PiCCO measurements concerning CO, EVLW, and GEDV.
The additional results of the PiCCO measurements, GEDVI, and CI are shown in Table 2. There were no differences between the CI values in the different age categories. However, the GEDVI values of the youngest age group were lower compared with the oldest group (p < 0.05). The median value of EVLWIh for the entire cohort was 200 mL/m (5th-95th percentile, 125-315 mL/m).

DISCUSSION
Our study confirms that young children have higher values of body weight-indexed lung water compared with older children or adults. The collected data may possibly be used as agerelated near-normal values for EVLWI.
High levels of lung water in young children have been described in the past and the present study is in conjunction with these results (4,11,12,19). However, these studies reflect critically ill children with predominant pulmonary abnormalities and frequently severe edema. This is the first study to describe measurements after these pulmonary abnormalities resolved, and therefore, the results can be regarded as near normal. Based on this study, the following values may represent near-normal indexed lung water when indexed to body weight. For children under 1 year old, EVLWI values in a range of 9-29 mL/kg might be considered near normal. For children between 1 and 5 years, near-normal values range from 7 to 25 mL/kg. Between 5 and 17 years, these values range from 5 to 13 mL/kg. Above 17 years, adult values (< 10 mL/kg) should be used.
In a former study, we explained the high lung water values in children by the altered relation between body weight and lung mass of children compared with adults (13). Also, there may be an age-related confounder in the calculation of EVLW. The relation between GEDV and intrathoracic blood volume is not constant in all ages and may vary even among individuals and species (4,20). Yet, the PiCCO system assumes a constant relationship and uses an adult-based calculation algorithm of EVLW values, thereby causing an erroneous estimation of EVLW (4,10,21,22).
In general, indexing requires a linear relation between the physiologic variable and the indexing variable over an age range from newborns to adults. The question is, which body variable is most suitable for indexing hemodynamic variables. Several adult studies have shown predicted body weight based on height to be superior over actual body weight in predicting severity of illness (14)(15)(16). For that reason, we also choose to index EVLW to height in our study. With regard to lung mass, it seems that height is the most suitable variable, as lung volume is significantly linear related to height (23). Recent studies in adults suggest that height seems to be superior to weight as an indexing variable (16,24,25). Our study results confirm that indexing to height might be more useful compared with the default indexing to actual body weight, which makes age-dependent adjustments unnecessary. The before-mentioned near-normal value categories based on weight might be  replaced by the more easy to use and age-independent nearnormal value of EVLWIh between 130 and 310 mL/m. In this study, CO indexed to BSA was age independent, which suggests the possibility of extrapolating adult normal values to children. In contrast, GEDVI appears to be age dependent and in accordance to an earlier study, lower in young children (13).
There are several limitations to our study. First, data were collected at various PICUs and with various PiCCO devices without a uniform treatment protocol. On the other hand, the PiCCO measurements were routinely performed as advised by the manufacturer. Furthermore, we collected absolute values to prevent indexing differences based on various algorithms incorporated in the subsequent software versions of the PiCCO devices. Second, the sample size is relatively small. This is specifically of importance for the children under 1 year of age group, as the values of EVLW are highest in this group. Third, the children were primarily selected based on a normal pulmonary condition and not on normal preload and CO conditions. Therefore, CI and GEDVI values from this study may not be entirely interpreted as normal. Fourth, we cannot rule out underlying pulmonary pathology, as no additional pulmonary examinations were included. On the other hand, the presented oxygenation and respiratory variables point to a near-normal pulmonary condition although some Pao 2 /Fio 2 ratios or A-a gradients were not normal. Therefore, the values of EVLWI should be interpreted as near normal. Fifth, some older children showed EVLWI levels above 10 mL/kg, thus according to adult reference values above normal. It might be possible that these patients, although without clinical signs of pulmonary edema, suffered from subclinical edema. However, a recent study in adults without pulmonary edema also showed higher levels of EVLWI (16). Therefore, we believe the higher levels in these children may be caused by the insufficient indexing system using actual body weight.
Obviously, it would be desirable to acquire data from healthy children. However, since lung water measurements require invasive arterial and central venous catheters, we see no opportunities to measure lung water in healthy children. Perioperative measurements (for instance, after induction of anesthesia and insertion of invasive catheters) might seem a reasonable alternative but would, with regard to the invasive catheters, concern major surgical procedures. In these cases, induction of anesthesia with concomitant fluid administration will influence hemodynamics and probably also lung water. Also all these children will be all endotracheally intubated and mechanically ventilated. Therefore, this group can also not be regarded as healthy normal controls. Since, at present, no other data are available, we believe our results are most suitable to use as near-normal values.

CONCLUSION
We conclude that in younger children, even after resolution of critical illness, values of actual body weight-indexed EVLW are higher compared with published adult values. We present agerelated near-normal EVLWI values. However, without further validation, these values can only be used with extreme caution. Furthermore, we advise to index EVLW to height, which seems to be age independent.

aCKNOWLEDGMENTS
We thank Khaled Menif, MD (PICU Tunis, Tunisia), and Harriet Adamson (Pulsion company) for their cooperation in gathering the patient data.