INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Since Bryan first advocated the prone position as a therapeutic maneuver in lung disease (1) numerous studies in both adult and pediatric subjects have almost uniformly reported an improvement in PA_O2 in the prone position compared to supine (2). The prone position has also been advocated as being beneficial for a wide variety of non-pulmonary considerations, ranging from improved neurodevelopment in neonates (7) to the promotion of flatulence (8). Although many authors hypothesize the improvement in oxygenation is due to an increase in functional residual capacity (FRC) (4, 5, 9) there are no published studies documenting positional changes in FRC in children with severe lung disease. A number of investigators have examined pulmonary mechanics in the prone and supine positions (6). However these studies have largely been carried out in spontaneously breathing neonates (some intubated and some not) with, at worst, mild lung disease. Although it may be advantageous to improve oxygenation and mechanics in this situation, we believe it is of greater importance to examine this phenomenon in patients with more severe lung disease, where marginal improvements in oxygenation and pulmonary function may have greater clinical impact. Additionally, patients with severe lung disease are generally under neuromuscular blockade, which affects the mechanics of the respiratory system (12), and we might therefore expect to observe different changes to those described by other investigators.

We therefore undertook this study to determine firstly whether FRC does increase in paralyzed, ventilated children when repositioned prone from supine, and secondly whether the expected improvement in PA_O2 could be attributed to any change in FRC. We report here the results in three groups of patients with relatively normal lung mechanics, severe restrictive lung disease, and severe obstructive disease.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients were eligible for enrollment if they were intubated wth cuffed endotracheal tubes (routine practice in our PICU) (13) and had no contraindication to being placed in the prone position (head or spine injury, facial surgery, abdominal distension). Informed consent was obtained from the parents of all patients enrolled in the study.

All measurements of pulmonary function were carried out with the endotracheal tube cuff inflated to prevent leaks at up to 40 cm H₂O, under neuromuscular blockade (vecuronium 0.1 mg/kg) with appropriate sedation (morphine or fentanyl, with diazepam or midazolam) and mechanically ventilated at settings determined by the attending pediatric intensivist. A computer controlled pulmonary function cart (Sensormedics 2600; Sensormedics, Yorba Linda, CA) was used to collect all data. Single breath static respiratory system resistance and compliance (R_rs, C_rs) were measured following a 200-300 ms end-inspiratory occlusion. Pressure was measured during the airway occlusion, and the flow-volume relationship of the subsequent passive expiration recorded, allowing both compliance and resistance to be calculated from a single breath (14). FRC was measured by nitrogen washout using Sivan's modification (15) of Gerhardt's technique (16). Arterial blood gases (ABG) were collected via indwelling arterial lines or peripheral arterial puncture, and analyzed immediately (Corning #178 pH / blood gas analyzer).

Classification into obstructive, restrictive and control groups was performed on the basis of clinical diagnosis and physical examination. Control patients were also required to have a normal CXR and respiratory compliance per kg of  0.80. In all patients, lung mechanics were consistent with the clinical classification.

In our PICU, patients are generally nursed in the supine position, usually with a 10-20 degree elevation of the head of the bed. For the purposes of this study, the bed was positioned horizontally for a minimum of 30 min before commencement of testing. Subjects were positioned supine with arms at their sides (i.e., not abducted). Patients were tested using clinically chosen ventilator settings. No attempt was made to standardize settings (e.g., PEEP) before lung function testing and ventilation was left unchanged for the duration of the study. Baseline tests were obtained with the patient in the supine position following routine suctioning of the endotracheal tube. Suctioning was subsequently performed as clinically indicated. During testing in the supine position the head was turned to one side in order to ensure a close approximation between ETT curvature in both supine and prone positions, as ETT curvature can affect measurement of R_rs (17). Tests were performed in the following order: ABG, FRC, C_rs, and R_rs. Following collection of baseline data, the subjects were placed in the prone position. The abdomen was not supported in any way. Arm position, which can affect FRC (18) was maintained as in the supine posture. The head was turned to one side as explained above. A second set of data was collected after 10 minutes in the prone position. Patients were then left undisturbed for 60 minutes before having further measurements of ABG and FRC. Following this data collection they were returned to the supine position where they once again had ABG and FRC measured after a 10 minute period. Vital signs were monitored throughout the study.

Statistical Analysis

Means are reported as mean ± SEM. Comparisons between patients in different positions were performed using one way analysis of variance (ANOVA). Paired samples where variables were measured only twice in each individual were compared with the Student's t-test. Relationships between continuous variables were described by simple linear regression. Both the correlation coefficient and the p value of the regression F test are reported. For all statistical analyses, a p value of < 0.05 was regarded as significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A total of 30 patients were studied. Ages ranged from 3 to 7.6 yr with a mean (± SEM) of 1.25 ± 0.30 yr. Patients had been ventilated for 2 h to 69 d (µ = 11.9 ± 3.3 d) prior to being studied and were prospectively classified as either normal, restictive, or obstructive based on diagnosis and physical examination and confirmed with baseline lung mechanics and volumes. There were significant differences in these parameters between the three groups, detailed in Table 1. In particular, the control subjects had significantly higher C_rs values and AaPO₂ ratios than either restrictive or obstructive groups, the restrictive group was distinguished by smaller FRC values than either control or obstructive groups and the obstructive patients demonstrated significantly higher R_rs than both control and restrictive groups. Neither mean duration of ventilation prior to testing or patient age differed significantly between the three groups. There was a tendency for patients in the obstructive group to be younger than the other groups, with a mean age of 0.35 ± 0.16 years compared to 2.08 ± 0.71 yr for controls and 1.31 ± 0.27 yr for restrictive disease, however this just failed to reach statistical significance (p = 0.054). The trend related to the majority of infants with obstructive lung disease suffering from RSV positive bronchiolitis.

PEEP in control subjects ranged from 2-4 cm H₂O (µ ± SEM = 3.8 ± 0.2), in obstructive patients from 2-6 cm H₂O (4.3 ± 0.4) and in patients with restrictive disease from 4-10 cm H₂O (5.8 ± 0.8). Mean PEEP in control subjects was significantly different to that of patients with restrictive disease (p = 0.02); no other differences between groups were noted.

Diagnoses were as followsin the control group: head injury (2), cerebral hemorrhage (1), meningitis (1), botulism (1), seizure disorder (1), treated cardiomyopathy (1), drug ingestion (1) upper airway obstruction (2); in the restrictive group: ARDS (4), pneumonia (4), pulmonary contusion (1), pulmonary edema (1); and in the obstructive group: RSV positive bronchiolitis (9, including one with BPD and one with VSD), and bronchiolitis not RSV positive (1). No clinically significant changes in vital signs were observed in the prone position, nor were any other complications encountered.

FRC measurements were reproducible with a mean coefficient of variation (CV) of 2.0 ± 0.4%; R_rs and C_rs were taken from a minimum of 10 breaths, with mean CV of 7.5 ± 0.9% and 4.7 ± 0.3%, respectively.

Individual patients showed a wide variation in FRC with position changes. Prone positioning at 10 min resulted in changes in FRC ranging from 13.2% (in a control patient) to +59.8% (in a patient with restrictive disease) of baseline (supine) FRC, however as a group these children showed no significant changes in FRC with changes in position. Although FRC tended to increase slightly in the prone position (mean of 22.2 ± 1.4 to 23.4 ± 1.5 ml/kg supine to prone) and subsequently decrease when moved from prone to supine one hour later (23.7 ± 1.5 to 22.5 ± 1.4 ml/kg) these changes were not statistically significant. The pattern of FRC increasing in the prone position and decreasing on return to the supine position was seen in all subgroups but again did not achieve statistical significance (Table 2).

Significant variation was also seen with AaPO₂; individual patients demonstrating changes with the initial move to prone positioning of between 28.6% and +42.6% of baseline values recorded in the supine position (both these extremes obtained in patients with restrictive disease; FRC in these patients altered by +17.0% and +19.7%, respectively). For the cohort of 30 patients AaPO₂ ratios improved from a mean of 0.44 ± 0.04 to 0.49 ± 0.04 when moving from the supine to the prone position. Once again, however, this change failed to reach statistical significance, with p = 0.075. Because some patients did not have arterial lines in situ, incomplete data was obtained for subsequent position changes. Subgroup analysis (Table 3) demonstrated significant changes only for the initial movement from supine to prone in the obstructive group, with the AaPO₂ ratio improving from 0.35 ± 0.03 to 0.38 ± 0.04 (p = 0.027).

Changes in AaPO₂ ratio could not be directly related to changes in FRC. Comparing fractional change in AaPO₂ with fractional change in FRC yielded a correlation coefficient r = 0.225, with p = 0.23.

Examination of changes in lung mechanics, summarized in Table 4, yielded one unexpected and previously unreported result in that prone positioning was associated with a significant improvement in R_rs in the group of patients with obstructive lung disease. R_rs decreased from 0.264 ± 0.024 to 0.216 ± 0.021 cm H₂O/ml/s with p = 0.0094 by paired Student's t-test. There was a trend to improvement in R_rs values in both the control and restrictive groups which did not reach statistical significance. Similarly, C_rs values were all marginally improved in the prone position but not by a significant degree.

Within the obstructive subgroup where significant changes in mechanics and oxygenation could be demonstrated the magnitude of these changes were not related to either patient age, or duration of mechanical ventilation at the time of testing.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although improved oxygenation in the prone position has been well documented in adult, pediatric and animal studies, the precise mechanism has yet to be elucidated. Several authors have proposed that an increase in lung volume might underly this phenomenon (4, 5, 9). FRC is known to increase in spontaneously breathing adult subjects in the prone position, in both awake (19) and anesthetized (18) states, and there are sound theoretical reasons why this should occur (1). To the best of our knowledge, changes in FRC with position have not been reported previously in critically ill children, although a study in spontaneously breathing intubated neonates with mild lung disease failed to demonstrate any increase in FRC despite significant increases in oxygenation (6) and a study in healthy neonates similarly reported no changes in FRC in prone versus supine positions (20). The combination of neuromuscular blockade and positive pressure ventilation may be expected to attenuate position-related changes in lung volumes and mechanics compared with those seen in spontaneously breathing subjects. Nevertheless, substantial improvements in oxygenation with prone positioning have been reported in both adult and pediatric patients under these conditions (2, 5). We undertook this study to determine firstly whether the FRC would increase in the prone position in children under neuromuscular blockade, and secondly whether the expected increase in oxygenation could be attributed to any change in FRC.

We elected to position the patients prone without any attempt to reduce pressure on the abdominal contents by the use of slings or supports as used in some other studies. (4, 6). This approach was chosen for simplicity, and because improved oxygenation has been extensively reported in pediatric and adult studies without the use of abdominal support.

Despite some impressive individual results, we were unable to demonstrate any significant change in FRC in any group of patients moved from supine to prone. Simple alterations in diaphragmatic mechanics should lead to an increase in FRC in the prone position, particularly in spontaneously breathing subjects. According to the Laplace law, for a given tension in the diaphragm, pressure and radius will vary inversely to one another. Thus, in the dorsal area of the diaphragm, which has a smaller radius of curvature than the ventral segments, a greater pressure will be exerted for any given wall tension. In the supine position, this increased pressure is opposed by the hydrostatic pressure of the abdominal contents which is greater dorsally than ventrally. However, in the prone position the increased pressure is allowed to act relatively unopposed, and thus at rest (FRC) the diaphragm will assume a position of increased caudal displacement, i.e., increased FRC. In patients under neuromuscular blockade, however, positive pressure ventilation will be distributed more evenly to both dorsal and ventral regions (1), lessening this effect. Nonetheless, pressure exerted by abdominal contents may still play a role, as in dorsal regions the more positive pleural pressure may result in a transpulmonary pressure less than the airway opening pressure, leading to regional atelectasis. This has been recently demonstrated in a canine model of oleic acid injury (21). Prone positioning allows a decrease in abdominal pressure in the dorsal regions with a corresponding increase in transpulmonary pressure, potentially allowing previously atelectatic segments to re-expand. This should translate to an increase in FRC, however the effect may be marginal and not detected by our testing. A third mechanism by which the prone position may lead to an increased FRC is through improved drainage of secretions. This may take time to manifest and our study was not designed to identify this potential effect.

Increase in oxygenation following prone positioning was seen only in the subgroup of patients with obstructive lung disease. Furthermore, while this achieved statistical significance, the clinical significance was minimal. Although individual patients demonstrated clinically important improvements (for example one patient with restrictive disease had an AaPO₂ ratio of 0.47 supine and 0.67 pronean improvement of 42.6%) the subgroup with severe restrictive disease and the control group did not demonstrate statistically significant position related changes in oxygenation. In clinical practice, oxygenation is most difficult to achieve in patients with restrictive lung disease, and it is therefore disappointing we could not demonstrate a consistent improvement in this subgroup.

Furthermore, we could not relate changes in FRC to changes in oxygenation. Despite the fact there were no substantial changes in either parameter in most of the subgroups, individual patients demonstrated significant swings in each parameter and we might still have expected to see a correlation between FRC and AaPO₂ if both had changed in the same direction in individual patients. This was not the case. In this respect, it is interesting to note that Richardson and Jung (22) showed that application of increasing amounts of PEEP in infants with hyaline membrane disease resulted in increases in FRC and PA_O2, but the two were not related. Improvements in oxygenation in the prone position have been demonstrated in animal studies to be due to a reduction of intrapulmonary shunting (21, 23). This reduction can result from redistribution of blood flow to better ventilated alveoli, or redistribution of ventilation to better perfused capillaries. CT scans in adult subjects performed within 10 min of a change from the supine to the prone position (i.e. similar timing to our study) demonstrate disappearance of posterobasal densities and appearance of new anterior densities in the prone position unrelated to any changes in PA_O2, Pa_CO2, cardiac index, right to left intrapulmonary shunt, or pulmonary arterial pressure (3, 5). The rapidity of the changes, together with the non-gravitational distribution of excess tissue mass excludes both resorption and development of edema (in newly non-dependent and dependent areas respectively), and free movement of interstitial fluid through the lungs.

There is evidence that pulmonary vascular resistance (PVR) is intrinsically lower in the dorsal lung regions (24, 25)i.e., that the increased regional flow to dorsal regions seen in the supine position is not solely due to gravitational forces but related to regional differences in PVR. Thus in the prone position, a more uniform distribution of blood is expected, as gravitational forces oppose rather than augment the differences in PVR. However, redistribution of pulmonary blood flow alone cannot explain the degree of change observed in CT studies (3, 5) and thus redistribution of intrapulmonary gas must also be involved. Redistribution of ventilation has been demonstrated as the principal mechanism of improved ventilation-perfusion matching in a recent study of oleic acid injured lungs in dogs (21).

The lack of improvement in oxygenation with prone positioning in our patients is at odds with other published data in animals, children and adults. One point of difference between our study and those of other authors is the duration of ventilation prior to repositioning. Animal studies investigate this phenomenon in recently injured lungs, for obvious reasons. Of the adult studies one (5) does not report duration of ventilation while others investigated patients ventilated for means of 2.1 d (4) and 3.2 d (3). The one neonatal study to report on patients with some (albeit mild) lung disease again investigated subjects relatively early in their course (mean duration of ventilation 5 d) (6). Murdoch and Storman (2) in a study of seven children with ARDS did not report duration of ventilation. It is of note that our patients were ventilated for relatively prolonged periods before being studied (for the cohort of 30, mean duration of ventilation was 11.9 d), and it is possible this may influence the degree of redistribution of ventilation and perfusion obtained with prone positioning. Arguing against this, of course, is the fact that the obstructive patients who did respond with improved oxygenation had a mean duration of ventilation of 15.2 d.

One unexpected finding in this study was the significant decrease of R_rs in patients with obstructive airway disease. R_rs does vary inversely with lung volume (26, 27), but FRC did not change significantly in this subgroup of patients. A decrease in R_L in the prone position was reported by Mendoza and colleagues (11), although most authors have found no change (6, 10). To the best of our knowledge this is the first study to measure positional changes in R_rs in patients with severe obstructive disease. It is possible that the prone position improves R_rs by relieving the gravity-mediated compression of large airways by mediastinal contents. The change observed in R_rs in obstructed patients was of greater magnitude than changes in any other parameter; and is perhaps the most clinically significant.

Potential risks of the prone position include accidental extubation, pressure sores including corneal abrasions, and traction injuries to the brachial plexus if the arms are abducted. No complications were seen in our group of patients who were left prone for only 1 hour. The position also presents difficulties for routine procedures such as the placement of indwelling vascular catheters, and would be potentially detrimental in patients requiring emergency resuscitation following unexpected cardiac arrest.

In 1974, Bryan clearly set out a sound theoretical basis for expecting an increase in lung volume in the prone position (1). This theory has not previously been put to the test in critically ill children in whom an increase in lung volume would presumably be most beneficial. We undertook this study in order to investigate the relationship between position, lung volumes, respiratory mechanics and oxygenation. Our findings indicate that in the setting of neuromuscular blockade and positive pressure ventilation positional changes in lung volumes do not occur.

We were unable to demonstrate any significant effects apart from a decrease in Rrs and an improvement in AaPO₂ in patients with obstructive disease.

Given the inconvenience and potential risks of prone positioning we do not recommend this position be used routinely in pediatric patients with the exception of patients with obstructive airways disease in whom the improvement in oxygenation and R_rs may be of benefit. This subgroup should be investigated further.