Objective. To determine whether a ventilatory strategy of permissive hypercapnia (PHC) reduces the duration of assisted ventilation in surfactant-treated neonates weighing 601 to 1250 g at birth.
Design. Forty-nine surfactant-treated preterm infants (birth weight: 854 ± 163 g; gestational age: 26 ± 1.4 weeks) receiving assisted ventilation were randomized during the first 24 hours of age to a PHC group (Paco 2: 45–55 mm Hg) or to a normocapnia group (NC; Paco 2: 35–45 mm Hg). The primary outcome measure was the total number of days on assisted ventilation. Uniform extubation and reintubation criteria were used for both groups. All patients received aminophylline before extubation.
Results. The total number of days on assisted ventilation expressed as median (25th–75th percentiles) was 2.5 (1.5–11.5) in the PHC group and 9.5 (2.0–22.5) in the NC group (Mann-Whitney U test). The number of patients on assisted ventilation throughout the first 96 hours after randomization was lower in the PHC group (log rank test). During that period, the ventilated patients in the PHC group had a higher Paco 2 and lower peak inspiratory pressure, mean airway pressure, and ventilator rate than did those in the NC group. The percentage of patients requiring reintubation within 24 hours postextubation (PHC 17% vs NC 28%) and supplemental oxygen at 28 days of life (PHC 43% vs NC 64%) and the total days of oxygen supplementation (PHC 15 [4–53] vs NC 32 [17–50]) did not differ between the groups. There were no differences in mortality, air leaks, intraventricular hemorrhage, periventricular leukomalacia, retinopathy of prematurity, or patent ductus arteriosus.
Conclusion. A ventilatory strategy of PHC in preterm infants who receive assisted ventilation is feasible, seems safe, and may reduce the duration of assisted ventilation. assisted ventilation, respiratory distress syndrome, gentle ventilation, lung injury.
- BPD =
- bronchopulmonary dysplasia •
- CLD =
- chronic lung disease •
- PHC =
- permissive hypercapnia •
- RDS =
- respiratory distress syndrome •
- NC =
Despite important advances in respiratory care, lung injury is still a leading cause of neonatal morbidity in neonates who receive ventilatory support. Although surfactant administration in neonates is very effective, it has not decreased the incidence of bronchopulmonary dysplasia/chronic lung disease (BPD/CLD).1 ,2 The duration and intensity of assisted ventilation may be important determinants in the development of lung injury. Large tidal volumes are considered particularly harmful to the developing lung.3Retrospective studies suggest that ventilatory strategies that lead to relative hypocapnia during the first days after birth may increase BPD/CLD.4 ,5 Permissive hypercapnia (PHC) is a strategy for the management of patients receiving assisted ventilation in which relatively high levels of Paco 2 are accepted to avoid high tidal volumes, pulmonary overdistention, and hypocapnia, thus potentially reducing lung injury.6 ,7 PHC has been reported to be a safe alternative to conventional normocapnic assisted ventilation in adult patients with acute respiratory distress syndrome8–10 and may increase survival.11PHC may be important in the management of preterm infants with respiratory distress syndrome (RDS).12 Data on hypercapnia in ventilated neonates are limited, and none of the reported studies predetermined or controlled the degree of hypercapnia. We performed a randomized, controlled pilot study to evaluate whether a strategy of PHC, initiated during the first 24 hours after birth in neonates weighing 601 to 1250 g at birth, decreases the number of days of assisted ventilation.
The primary hypothesis was that a ventilatory strategy of PHC would reduce the number of days on assisted ventilation by 35%, compared with a normocapnic approach. The primary outcome measure was the total number of days on assisted ventilation. The secondary outcome measures included the total number of days on supplemental oxygen, the incidence of BPD/CLD, postnatal steroid administration, air leaks, intraventricular hemorrhage, periventricular leukomalacia, retinopathy of prematurity, patent ductus arteriosus, necrotizing enterocolitis (≥stage II-A), and length of hospitalization.
Eligibility and Randomization
This study was conducted in the Regional Newborn Intensive Care Unit at the University of Alabama at Birmingham between November 1, 1995 and December 9, 1996. The study protocol and informed consent form were approved by the institutional review board.
Infants were eligible for the study if all the following criteria were met: 1) birth weight of 601 to 1250 g; 2) surfactant-treated RDS on assisted ventilation; 3) postnatal age <24 hours; and 4) written parental informed consent. Infants were excluded for any of the following reasons: 1) 5-minute Apgar score <3; 2) small for gestational age; 3) congenital anomalies or suspected congenital infection; 4) multiple pregnancy of triplets or more; and 5) infant not expected to need prolonged ventilatory assistance as judged by the attending neonatologist. The patients were assigned to either a PHC or a normocapnia (NC) group using a permuted block randomization procedure consisting of a random sequence of blocks of 4, 6, 8, and 10. The group assignments were recorded and sealed within sequentially numbered opaque envelopes. The odds of assignment to one of the two groups were not known to the investigators. To ensure a similar birth weight mix, infants were stratified into the following groups: 601 to 750 g, 751 to 1000 g, and 1001 to 1250 g. Twins who met enrollment criteria were randomized to the same group according to the randomization of the first eligible infant of the pair.
All infants received initial assisted ventilation with an Infant Star ventilator (Infrasonics, Inc, San Diego, CA), a time-cycled pressure-limited ventilator. High frequency ventilation was used at the discretion of the attending physician. Oxygenation was monitored continuously with a pulse oximeter. Ventilatory management was the responsibility of the attending physicians, but different algorithms were provided for infants randomized to each treatment group. The goals for pH and Paco 2 were different in the two groups of infants. In the NC group, the goals were to keep Paco 2 between 35 and 45 mm Hg and pH ≥7.25. In the PHC group, ventilatory management was directed to maintain Paco 2 between 45 and 55 mm Hg and pH ≥7.20. These goals were used for the first 96 hours after randomization. After that time, the changes in the ventilator settings were directed at the pH criteria, allowing high levels of Paco 2 also in the NC group. The goal for Pao 2 level was between 50 and 80 mm Hg in both groups.
To achieve these goals, ventilator setting changes followed a modified clinical algorithm, developed for the management of pressure-limited ventilation in infants with RDS.13 The algorithm incorporates accepted concepts of the effects of neonatal assisted ventilation on gas exchange. Briefly, CO2elimination was enhanced by increasing the ventilatory frequency. If this failed, peak inspiratory pressure was increased. During the weaning process, priority was given to reduce peak inspiratory pressure. Oxygenation was improved predominantly by increases in mean airway pressure when the Fio 2 was >0.70. This was accomplished by increasing positive end expiratory pressure or inspiratory time. If hypoxemia was persistent, peak inspiratory pressure was increased. If Fio 2 was <0.40, increases in oxygen concentration were performed for hypoxemia. When the Fio 2 was between 0.40 and 0.70, the parameter modified was decided based on the level of mean airway pressure and chest excursion. Although the algorithm was based on arterial blood gas measurements, clinical assessment, including, but not limited to, chest wall movements, breath sounds, and cardiovascular function, was performed simultaneously. The algorithm suggested the type of ventilator setting change but not its magnitude, which was decided on clinical grounds.
Objective criteria were used for extubation to minimize bias. Infants were extubated from assisted ventilation if all the following criteria were met: peak inspiratory pressure ≤19 cm H2O, ventilator rate ≤10 per minute, Fio 2≤0.4, and arterial pH ≥7.25. An aminophylline loading dose was given before extubation. Continuous positive airway pressure was used as clinically indicated. Reintubation was performed for a pH <7.20, respiratory failure, or severe apneic episodes needing assisted ventilation according to the attending physician. The defined extubation criteria were followed for every period on assisted ventilation, except when patients required more than one reintubation for apnea. In these patients, a new extubation was attempted 5 to 7 days after the previous failure. Patients were weaned from oxygen supplementation when they were able to maintain oxygen saturation ≥90% while breathing air.
At least one dose of surfactant (Survanta, Ross Laboratories, Columbus, OH) was administered before randomization. Repeated doses were given if the Fio 2 was >0.3 and/or if the mean airway pressure was >7 cm H2O. Dexamethasone was considered for infants who were ventilator-dependent at 10 days of age, if they had radiologic findings consistent with developing chronic lung injury in the absence of confounding circumstances, such as patent ductus arteriosus, pneumonia, or sepsis. A 7-day course of dexamethasone at 0.6 mg/kg per day was used. A second, tapering 9-day course (0.6, 0.4, and 0.2 mg/kg per day for 3 days each) was given as clinically indicated.
Intravenous fluids were started at 130 to 200 mL/kg per day in infants ≤750 g, and at 100 to 150 mL/kg per day in infants between 751 and 1250 g. The total fluid intake was adjusted as needed, based on urinary output, weight change, and serum sodium values. All patients received indomethacin for intraventricular hemorrhage prophylaxis (a daily dose of 0.1 mg/kg for 3 days). Sodium bicarbonate was given for serum bicarbonate ≤ 16 mEq/L.
Data Collection and Definitions
Maternal demographics and obstetric history of pregnancy, labor, and delivery were obtained from the mother's medical record. Gestational age was assessed by the best obstetrical estimate and the Ballard examination of the neonate. Infants whose birth weight was below two standard deviations were considered small for gestational age.14 An initial cranial ultrasound was performed before study entry. Subsequent studies were performed routinely on days 5 through 7 and 28 ± 7 or when clinically indicated. The arterial blood gas values were recorded before randomization and closest to 6, 12, 24, 36, 48, 60, 72, 84, and 96 hours postrandomization while the infant received assisted ventilation. The corresponding ventilator settings at these times were recorded. Data were collected at prospectively scheduled times and recorded on standardized forms.
The total duration of assisted ventilation was calculated from the sum of all periods of assisted ventilation until final extubation. Time on continuous positive airway pressure was not counted as assisted ventilation. The total duration of oxygen supplementation was calculated from the sum of all periods of any technique of oxygen supplementation, including after transfer or discharge. BPD was defined as oxygen requirement and abnormal chest radiograph on day 28 of postnatal age, with oxygen requirement for at least 21 of the first 28 days.15 Air leaks included pneumothorax and/or pulmonary interstitial emphysema. The severity of intraventricular hemorrhage was graded according to the criteria of Papile et al.16 A hemorrhage was considered to have progressed if: 1) a new intraventricular hemorrhage developed from an initial negative head ultrasound; 2) there was a progression in any grade of intraventricular hemorrhage; or 3) a second intraventricular hemorrhage was noted in the hemisphere opposite from the existing hemorrhage. A diagnosis of periventricular leukomalacia was made if the cranial ultrasound showed postnatal development of multiple cystic echolucencies in the cerebral white matter. Proven sepsis was defined as a positive blood culture result for bacteria or fungus treated by the clinicians at any time during hospitalization. The presence of patent ductus arteriosus was confirmed by echocardiography. Modified Bell's criteria were used for necrotizing enterocolitis staging.
Sample Size Determination and Statistical Methods
At the time when the study was designed, the mean number of days on assisted ventilation for infants weighing 601 to 1250 g in our unit was 14 days. Using an estimated standard deviation of 6 days, a sample size of 24 infants per group would be required to detect a 5-day difference (a reduction of 35%) in days on assisted ventilation between the groups with a power of 0.80 and a type I error of 0.05 (two-tailed). The analysis was performed according to intention to treat, and crossover was not allowed. Results were analyzed by the Student's t test, χ2, or Fisher's exact test, as appropriate for parametric data and by Mann-WhitneyU rank sum test for nonparametric data. A Kaplan-Meier analysis was performed, comparing the time on assisted ventilation during the first 96 hours after randomization and throughout the hospital stay in the two groups. A P value <0.05 was considered to be statistically significant.
During the enrollment period, a total of 171 preterm infants weighing 601 to 1250 g at birth were born at the University of Alabama at Birmingham (Fig 1). Of these infants, 57 did not receive assisted ventilation in the first 24 hours after birth; thus, 114 infants were potentially eligible. A total of 65 neonates were excluded for the following reasons: Apgar score <3 at 5 minutes (n = 7); small for gestational age (n = 5); congenital anomalies/infection (n = 11); triplet (n = 1); not expected to need prolonged mechanical ventilation (n = 24); consent denied (n = 12); or the attending physician's decision not to randomize based on evidence of pulmonary hypertension or early pulmonary interstitial emphysema (n = 5). Of the infants, 49 were randomly assigned, 24 to the PHC group and 25 to the NC group. All patients had radiographs consistent with RDS. Birth weight, gestational age, gender, race, prenatal steroid administration, maternal chorioamnionitis, percentage of cesarean section, postnatal age at study entry, Apgar scores, and prerandomization arterial blood gases and oxygenation indexes did not differ significantly between the two groups (Table 1). In addition, the degree and need for resuscitation, the overall delivery room management, and time to intubation did not differ between the groups.
The duration of assisted ventilation was not reduced significantly in the PHC group (Table 2). Because the results were not distributed normally, the median and the 25th–75th percentiles are used to express these values. The total number of days on assisted ventilation was 2.5 (1.5–11.5) in the PHC group and 9.5 (2.0–22.5) in the NC group (P = .17; Mann-WhitneyU test). However, the number of patients on assisted ventilation during the 96 hours after randomization was lower (P < .005; 1og rank test) in the PHC group (Fig 2).
During the first 96 hours after randomization, the ventilated patients in the PHC group had higher (P < .05) Paco 2 values (Fig 3) and lower peak inspiratory pressures, mean airway pressures, and ventilator rates than did those in the NC group at most 12 hour intervals after randomization (Fig 4); impairment of oxygenation, reflected by arterial to alveolar oxygen ratio and Fio 2, was comparable in the two groups (Fig 5).
The total duration of oxygen supplementation, the incidence of BPD, oxygen requirement at 36 weeks, air leaks, reintubations, and the use of postnatal steroids and postextubation continuous positive airway pressure did not differ significantly between the two groups (Table 2). The rate of reintubation within 24 hours postextubation was 17% in the PHC group and 28% in the NC group (not significant). The mean peak inspiratory pressure before extubation was 15 cm H2O in the PHC group and 16 cm H2O in the NC group (not significant). Of the survivors in the PHC group, 67% (14/21) required at least one reintubation at some point during their hospitalization (7 for respiratory failure, 5 for apnea, 1 for necrotizing enterocolitis, and 1 for ventriculo-peritoneal shunt placement). Of the survivors in the NC group, 55% (12/22) were reintubated (8 for respiratory failure, 3 for apnea, and 1 for necrotizing enterocolitis). Only 2 infants (both randomized to the NC group) received high-frequency ventilation at any time during the hospitalization.
Three patients in each group died. The causes of death in the PHC group were grade 4 intraventricular hemorrhage in 2 patients and necrotizing enterocolitis/sepsis in 1 patient. In the NC group, 2 patients died from grade 4 intraventricular hemorrhage and 1 died because of respiratory failure. There were no significant differences in any of the nonrespiratory outcome measures between the two groups (Table 3). Three patients (all in the PHC group) did not have a head ultrasound performed before randomization.
Assisted ventilation in infants with RDS has resulted in improved neonatal survival.17 However, its excessive or prolonged use has been associated with an increased incidence of BPD/CLD.18–20 In this pilot study, we found that a ventilatory strategy of PHC for the management of preterm infants on assisted ventilation is feasible and not associated with an increased rate of complications, compared with a traditional normocapnic approach. The duration of assisted ventilation was decreased in the first 96 hours after randomization in the PHC group, but the total duration of assisted ventilation in the PHC group did not decrease significantly.
Our study has some limitations that preclude firm conclusions. The failure to attain a level of statistical significance in the total duration of assisted ventilation may be primarily attributable to the marked variability in duration of ventilation and a small sample size, with the possibility of a type II error, because the power of the test was below the desired power of 0.80. The variability in the duration of ventilation was in part because a ventilatory strategy of PHC may not necessarily decrease the need for assisted ventilation in infants with apnea or other nonrespiratory causes. A choice of ranges of Paco 2 that separated the groups more may have led to larger differences in the duration of ventilation. Because this study could not have been masked, to decrease the influence of any potential bias on duration of assisted ventilation, we defined and followed strict extubation and reintubation criteria and used precise indications for those therapies that have been reported to influence extubation success, such as aminophylline17 and dexamethasone.21
PHC was compared with NC during the first 96 hours based on two reasons. First, the available evidence relates BPD/CLD to aggressive ventilation in the early (<96 hours) course of respiratory disease.4 ,5 Second, after the first few days of treatment, it is common practice in our unit to allow Paco 2 levels to rise, so a consensus was reached among the neonatologists to allow Paco 2 increases in the NC group also after the first 96 hours. The range of safe levels of Paco 2 in neonates is unknown. The benefits of PHC may be counterbalanced by the potential adverse consequences of mild hypercapnic acidosis. Severe hypercapnic acidosis increases cerebral blood flow, which may lead to cerebral edema, increased intracranial pressure, and intraventricular hemorrhage.22 ,23 However, data collected retrospectively show an association between intraventricular hemorrhage and severe (Paco 2: ≥60 mm Hg) but not mild hypercapnia.24 ,25 Paco 2levels >60 mm Hg also have been associated with the development of retinopathy of prematurity in a mouse model.26 On the other hand, recent clinical studies suggest that hypocapnia increases the risk for periventricular leukomalacia and cerebral palsy.27–29 Indeed, it is possible that mild hypercapnia may be protective to the immature brain.30 Based on this information and on the traditional thought that normal Paco 2 levels in ventilated infants are ∼40 mm Hg, the target Paco 2ranges selected for this pilot trial were considered safe for both groups. However, tolerance of even higher Paco 2 values and/or lower pH values in the PHC group may be acceptable and may have resulted in a larger effect size. Even when the treating physicians followed the protocol, the magnitude of the ventilator changes were sometimes insufficient and may have accounted for the lower than expected levels of Paco 2 in the PHC group. Increased spontaneous minute ventilation also may have limited the development of hypercapnia.
The lack of tidal volume measurements in this study may be criticized, because it has been recommended that PHC should be achieved with a reduction in tidal volume. Because tidal volume was not measured in all patients, we cannot relate our results to the tidal volumes used. Experimental and clinical data indicate that ventilator-induced lung injury contributes markedly to lung damage in immature lungs and that the deleterious effect of assisted ventilation is more related to the volumes than to the pressures administered.31 Thus, the term volutrauma is preferred by some over the commonly used term barotrauma.32 Tidal volume may be a sensitive indicator of the relative risk of ventilator-induced lung injury. However, tidal volumes may vary markedly during pressure-limited ventilation. Furthermore, the uniform practice of Paco 2 analysis and the infrequent routine use of tidal volume measurements in neonates make our approach more generalizable.
PHC has not been evaluated previously in a controlled study in neonates. A low mortality rate with the use of PHC has been reported in children33 and adults8–11 with acute respiratory distress syndrome and in neonates with persistent pulmonary hypertension.34 A report of 151 patients weighing ≤1500 g, managed with a protocol emphasizing minimization of ventilatory support, suggested that the use of low ventilator settings may reduce the incidence of pneumothorax and lung injury in neonates who receive assisted ventilation.35 Two large retrospective studies, designed to determine risk factors for BPD/CLD, concur on the importance of ventilatory strategies.4 ,5 Using multiple logistic regression analysis, these two studies independently concluded that ventilatory strategies that lead to hypocapnia during the early neonatal course result in an increased risk of BPD/CLD (relative risk: 1.45; 95% confidence interval: 1.04–2.01; odds ratio: 3.3; 95% confidence interval: 1.3–8.3). Kraybill et al4 performed a multicenter analysis in 235 infants with birth weights between 751 and 1000 g admitted to 10 neonatal units. In this study, only low levels of Paco 2 on days 2 and 4 of life and male gender were independent predictors of BPD (defined as receiving oxygen at 30 days of life). With a similar design, Garland et al5 analyzed data on 188 patients weighing <1700 g at birth. They observed that low levels of Paco 2 on the first day of life remained associated with CLD (defined as receiving oxygen at 36 weeks' gestational age) even when several typical measures of respiratory illness severity were put into the model. Peak levels of Paco 2 >50 mm Hg in the first 4 days of life and Paco 2 ≥41 mm Hg before the administration of surfactant were reported to be associated with a lower incidence of BPD/CLD in these two studies, respectively. The lack of a statistically significant reduction in the incidence of BPD in the current pilot study may be attributable to a type II error caused by the relatively small sample size.
Hypercapnia has physiologic effects on gas exchange that should provide important benefits. The ventilatory requirements in patients with respiratory failure are determined by the targeted Paco 2.36 ,37 If Paco 2 is permitted to increase, alveolar CO2 will increase. The increase in alveolar CO2 that occurs during PHC increases CO2 elimination for the same minute ventilation. Thus, as CO2 equilibrates at a higher level in the body, alveolar ventilation requirements decrease because the higher alveolar CO2 makes elimination more effective. Furthermore, for a given Pao 2 the shift to the right of the oxygen dissociation curve during hypercapnia (Bohr effect) permits more unloading of oxygen to the tissues. Cardiac output may improve as a result of the decreased mean airway pressure that can be used during PHC. The possible negative effects of PHC, thought to be of less clinical importance, include, among others, a small reduction in Pao 2 as a result of the increased alveolar CO2, a reduction in the transported oxygen in the arterial blood as a result of the right shift of the O2 dissociation curve, and the potential increases in work of breathing, pulmonary vascular resistance, and cerebral blood flow.6 ,7 ,9 ,10 ,36 ,37
PHC has been recommended as the preferred ventilatory strategy for the management of acute respiratory distress syndrome,37 although the data to support its use are limited. The data on potential benefits of PHC in neonates come from retrospective studies. Despite the extraordinarily common use of assisted ventilation in neonates, few prospective, randomized trials have been conducted to determine strategies that limit the severity or prevent the development of BPD/CLD in neonates requiring assisted ventilation. The results of this pilot study provide additional evidence that PHC is likely to be a safe alternative to the traditional normocapnic approach and may reduce lung injury and the duration of assisted ventilation. Moreover, PHC may be combined with other promising techniques such as high frequency ventilation17and tracheal gas insufflation.38 However, caution must be exercised before widespread use of PHC as a mode of assisted ventilation in preterm infants. The small sample size, the relatively low incidence of some of the outcome measures and adverse events, and the wide confidence intervals of the treatment effects preclude firm conclusions about potential benefits or adverse effects of PHC. Additional clinical trials to evaluate the effect of PHC in neonates are warranted.
This work was supported in part by Grant M01-RR00032 from the National Institutes of Health.
We thank the respiratory therapists, nurses, residents, fellows, and attending physicians whose collaboration made this trial possible. We also thank Cassandra Hudson and Maria G. Peroni, who helped us in the preparation of the manuscript.
- Received June 30, 1998.
- Accepted March 2, 1999.
Reprint requests to (W.A.C.) University of Alabama at Birmingham, 525 New Hillman Building, 619 S 19th St, Birmingham, AL 35233-7335. E-mail:
This work was presented at the Annual Meeting of the Society of Pediatric Research, Pediatric Academic Society; May 2–6, 1997; Washington, DC.
- ↵Soll RF. Surfactant therapy in the USA: trial and current routines. Biol Neonate. 1997;71:1–7. Supplement
- ↵Halliday HL. Clinical trials of surfactant replacement in Europe. Biol Neonate. 1997;71:8–12. Supplement
- Hernandez LA,
- Peevy KJ,
- Moise AA,
- Parker JC
- Carlo WA,
- Pacifico L,
- Chatburn RL,
- Fanaroff AA
- Ehrenkranz RA,
- Verter J,
- Fanaroff AA,
- et al.
- ↵Bancalari E, Sinclair JC, eds. Mechanical ventilation. In: Effective Care of the Newborn Infant. 1992:200
- Avery ME,
- Tooley WH,
- Keller JB,
- et al.
- Poets CF,
- Sens B
- Szymonowicz W,
- Yu VYH,
- Wilson FE
- Levene MI,
- Fawer CL,
- Lamont RF
- Wesolowski E,
- Smith LEH
- Graziani L,
- Spitzer AR,
- Mitchell DG,
- et al.
- Wiswell TE,
- Graziani LJ,
- Kornhauser MS,
- et al.
- Vannucci RC,
- Towfighi J,
- Heitjan DF,
- Brucklacher RM
- Botero C,
- Reda Z,
- Medoza P,
- Davis A,
- Harrison R
- Wung JT,
- James LS,
- Kilchevsky E,
- James E
- ↵Mariani GL, Carlo WA. Ventilatory management in neonates: science or art? In: Goldsmith JP, Spitzer AR, eds. Controversies in Neonatal Pulmonary Care. Philadelphia, PA: WB Saunders; 1998:33–48
- Copyright © 1999 American Academy of Pediatrics