PEDIATRICS Vol. 108 No. 1 July 2001, pp. 212-214

High-Frequency Oscillation and Chronic Lung Disease in Very Low Birth Weight Infants

To the Editor.

The study by Rimensberger et al¹ suggests that early institution of high-frequency oscillation (HFO) for the treatment of respiratory distress syndrome improves the pulmonary outcome in ventilated very low birth weight infants. Unfortunately, there are flaws in study design that limit the validity of this conclusion.

First, the 2 study cohorts were not concurrent. Most of the infants in the experimental cohort treated with HFO were born in 1997 and 1998, whereas those in the historical control cohort treated with conventional mechanical ventilation were born in 1994 and 1995. Subtle differences in clinical management might have been unavoidable in the 2-year period that separates these cohorts and contributed to the difference in the pulmonary outcome. Second, the statistical basis for selecting the specific number of infants in each cohort is not provided. Without appropriate power analysis, the statistical validity of comparisons between cohorts is difficult to confirm. Third, the retrospective nature of the study raises questions about uncontrolled data collection and problems with data retrieval, both inherent problems of a retrospective study design. Fourth, what is most striking is that there is no discussion of the common complications, such as symptomatic patent ductus arteriosus and airway infection, both of which are strong contributors to the development of chronic lung disease. And fifth, there is no discussion of the nutritional management of the infants. Nutrition plays a big role in the process of healing from lung injury.

Thus, although the zero incidence of chronic lung disease in very low birth weight infants ventilated with HFO appears to be a remarkable achievement indeed, the enthusiasm for accepting this conclusion is dampened by the flawed study design. The subtle message that a single ventilation strategy can eliminate chronic lung disease, a disease of diverse causes, is misleading. What is needed to address the important question of ideal ventilation is a prospective, randomized trial with a large number of infants and carefully controlled confounding variables.

Jayant P. Shenai, MD
Department of Pediatrics
Vanderbilt University Medical Center
Nashville, TN 37232

REFERENCE

1. Rimensberger PC, Beghetti M, Hanquinet S, Berner M First intention high-frequency oscillation with early lung volume optimization improves pulmonary outcome in very low birth weight infants with respiratory distress syndrome. Pediatrics. 2000; 105:1202-1208 [Abstract/Free Full Text]

In Reply.

Dr Shenai focusses his concerns on a statement never made in our publicationnamely, that a single ventilation strategy can eliminate chronic lung disease. We are well aware, as pointed out in the "Discussion" section, that using high-frequency oscillation (HFO) as the first and exclusive mode of mechanical ventilation in premature infants at high risk to develop chronic lung disease (CLD) may not be the only possible approach to reduce ventilator-associated lung lesions in this group of patients. What we would like to stress are the following 2 concepts. The first is that ventilator-induced lung injury can occur rapidly and starts in the delivery room. Therefore, a lung protective strategy has to be used from the beginning, if at all possible from the first ventilator breath, as experimental data strongly suggests.^1,2 The second concept is that an open lung approach (open up the lung and keep it open) is essential to the avoidance of acute and chronic lung injury.^3-8 We clearly stated that the use of HFO may not be the only way to accomplish these goals, and we believe that identical results might be achieved with conventional ventilation using an open lung approach.⁹ Therefore, we certainly agree with Dr Shenai that a prospective, randomized trial with large numbers of infants will be required to address the important question of ideal ventilation.

We clearly stated that our study¹ was a retrospective review looking at outcome data in a well-defined population (consecutively admitted patients over a defined period of time for both groupsthe HFO group and the historical control group) involving every single patient meeting the eligibility criteria (ie, <32 weeks of gestational age and <1500 g birth weight) being ventilated exclusively with one or the other mode. This allowed us to have 2 relatively homogeneous cohorts of patients without selection bias. Furthermore, all patients were treated otherwise by the identical medical staff and with unchanged treatment strategies with the exception of the ventilator mode, although subtle changes in clinical management cannot be excluded totally. To do posthoc power calculations to underlie the significance of results from a retrospective observational study comparing 2 cohorts is neither appropriate nor recommended by statisticians. We are well aware that our study does not have the power of a randomized, controlled trial. However, using calculations of the extremes of the 95% confidence interval (CI) to estimate the maximal frequency of events (ie, CLD at 36 weeks' postconceptional age) allows an estimation of the possible power of our results. These calculations of the exact CI show a possible frequency of CLD of 0% to 11% for the HFO group, and 19% to 51% for the conventional mechanical ventilation (CMV) group. Depending on the chosen values, a randomized, controlled trial will need 66 patients to detect an incidence of CLD of 5% (HFO group) vs 35% (CMV group), 176 patients (5% vs 20%), and 440 patients (10% vs 20%) with a power of 80% and an alpha of 5% (bilateral) as significantly different. The difference of 5% vs 35% might be optimistic but corresponds to our observational results, and the number of patients required fits almost our patient numbers.

Regarding the question raised about common complications, there was no difference in the incidence of a patent ductus arteriosus requiring closure (19 (70%) in the HFO group and 23 (66%) in the control group, respectively: P = .78, Fisher's exact test). All patients with proven congenital pneumonia were excluded from analysis. Only 3 of 27 patients in the HFO group and 2 of 34 patients in the CMV group were treated with a full course of antibiotics within the first 2 weeks of life for proven sepsis.

We are well aware that our experience cannot be generalized, and we would not recommend using our strategy of "first intention HFO with early lung volume optimization" in units with no experience in HFO, without waiting for additional confirmation of our observational data by a randomized, controlled trial.

Peter Rimensberger, MD
Pediatric and Neonatal Intensive Care Unit
Hopital des Enfants
University Hospital of Geneva
CH-1211 Geneva 14, Switzerland

REFERENCES

1. Meredith KS, deLemos RA, Coalson JJ, Role of lung injury in the pathogenesis of hyaline membrane disease in premature baboons. J Appl Physiol. 1989; 66:2150-2158 [Abstract/Free Full Text]
2. Jackson JC, Truog WE, Standaert TA, Effect of high-frequency ventilation on the development of alveolar edema in premature monkeys at risk for hyaline membrane disease. Am Rev Respir Dis. 1991; 143:865-871 [Medline]
3. McCulloch PR, Forkert PG, Froese AB Lung volume maintenance prevents lung injury during high frequency oscillatory ventilation in surfactant-deficient rabbits. Am Rev Respir Dis. 1988; 137:1185-1192 [Medline]
4. Muscedere JG, Mullen JBM, Gan K, Slutsky AS Tidal ventilation at low airway pressures can augment lung injury. Am J Respir Crit Care Med. 1994; 149:1327-1334 [Abstract]
5. Jackson JC, Truog WE, Standaert TA, Reduction in lung injury after combined surfactant and high-frequency ventilation. Am J Respir Crit Care Med. 1994; 150:534-539 [Abstract]
6. Clark RH, Gerstmann DR, Null DMJ, deLemos RA Prospective randomized comparison of high-frequency oscillatory and conventional ventilation in respiratory distress syndrome. Pediatrics. 1992; 89:5-12 [Abstract/Free Full Text]
7. Gerstmann DR, Minton SD, Stoddard RA, The Provo multicentre early high frequency oscillatory ventilation trial: improved pulmonary and clinical outcome in respiratory distress syndrome. Pediatrics. 1996; 98:1044-1057 [Abstract/Free Full Text]
8. Amato MB, Barbas CS, Medeiros DM, Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med. 1998; 338:347-354 [Abstract/Free Full Text]
9. Rimensberger PC, Pache JC, McKerlie C, Frndova H, Cox PN Lung recruitment and lung volume maintenance: a strategy for improving oxygenation and preventing lung injury during both conventional mechanical ventilation and high-frequency oscillation. Intensive Care Med. 2000; 26:745-755 [CrossRef][Medline]

In Reply.

We were surprised to find an article about high frequency oscillation (HFO)¹ using historical controls being published in Pediatrics, after a dozen randomized trials on high-frequency ventilation have become available in the literature. By publishing this study, Pediatrics sends a false message based on weak data to the caretakers of premature infants.

The danger of overly aggressive HFO lies in an increased incidence of air leaks² and intracranial hemorrhages (ICH). An increased incidence of severe ICH was found in at least 3 randomized trials.^3-5 It was also found in the study of Rimensberger et al (22% versus 13%), although not significant due to the small sample size, but worrisome in light of the declining incidence of ICH worldwide.

The reduction of chronic lung disease (CLD) in the study of Rimensberger et al¹ may be explained by 3 facts other than the use of HFO: first, it goes well along with the worldwide trend and thus may be easily explained by the use of historical controls. Second, withdrawal of support was used in as much as 16% of the cases in the HFO group compared with only 10% in the historical control group, which most likely reduced the incidence of CLD in the HFO group. Third, the incidence of CLD in the historic control group was very high, making it easy for the HFO group to look good. For these reasons, this study has a low level of evidence. In the discussion, Rimensberger et al¹ suggest that the outcome in their HFO group was improved because of limited exposure of the infants to conventional ventilation. However, their patients still received about 15 minutes of hand bagging, which is not substantially less than the median 28 minutes of conventional ventilation in our recent randomized trial.² Other possible reasons for the better outcome in their HFO group¹ in comparison with the HFO- treated infants in our recent randomized trial,² which are unfortunately not discussed by the authors, are the frequent withdrawal of support, thus preventing the survival of the highest risk infants, and higher birth weight and gestational age (mean gestational age 27.7 weeks vs 26.9 weeks, mean birth weight 981 g vs 913 g). Also, reporting only mean ± SD does not tell us whether infants with a gestational age below 25 weeks, ie, those with the highest risk for developing CLD, were included. There were 36 such infants in our study.² On page 1207, Rimensberger et al¹ falsely imply that 50% of our patients² were crossed to the alternate ventilation mode. In fact, it was only 13% (data not published).

When we planned our own randomized trial,² we conducted a similar retrospective analysis and found similarly huge benefits associated with the introduction of HFO in our nursery. We are glad that we did not publish this, because these benefits could not be found in our later randomized trial.² This confirmed once more the limited value of retrospective analyses.

In short, the differences between the HFO patients and the historical controls in this analysis, as well as the differences to the results of other randomized trials, may be easily explained by reasons unrelated to the ventilation mode.

Ulrich Thome, MD
Division of Neonatology
Department of Pediatrics
University of Alabama at Birmingham
Birmingham, AL 35233-7335

F. Pohlandt
Division of Neonatology
University Children's Hospital
89070 Ulm, Germany

REFERENCES

1. Rimensberger PC, Beghetti M, Hanquinet S, Berner M First intention high-frequency oscillation with early lung volume optimization improves pulmonary outcome in very low birth weight infants with respiratory distress syndrome. Pediatrics. 2000; 105:1202-1208
2. Thome U, Kössel H, Lipowsky G, Randomized comparison of high frequency ventilation with high rate intermittent positive pressure ventilation in preterm infants with respiratory failure. J Pediatr. 1999; 135:39-46 [CrossRef][Medline]
3. HIFI Study Group High-frequency oscillatory ventilation compared with conventional mechanical ventilation in the treatment of respiratory failure in preterm infants. N Engl J Med. 1989; 320:88-93 [Abstract]
4. HiFO Study Group Randomized study of high-frequency oscillatory ventilation in infants with severe respiratory distress syndrome. J Pediatr. 1993; 122:609-619 [Medline]
5. Moriette G, Walti H, Salanave B, Prospective randomized multicenter comparison of high-frequency oscillatory ventilation and conventional ventilation in preterm infants <30 weeks gestational age with RDS [abstract]. Pediatr Res. 1999; 45:212A

In Reply.

In response to the letter by Drs Thome and Pohlandt, we would like to discuss some of the concerns addressed by these authors concerning our publication.¹

Although the authors may have their personal opinion on what should be published and what not, we feel that it remains important that medical journals give the opportunity to publish new concepts (in our case an early strategy for prevention of ventilator-induced lung injury) to ensure progress within the medical community. We fully agree that the level of evidence of a randomized, controlled trial (RCT) is higher than the one by a study using historical controls, and that our observations will need confirmation by a well-designed RCT.

The issue on the risk of intracranial hemorrhages (ICH) started with the multicenter HIFI study² showing an increased risk for ICH in the group ventilated with high-frequency ventilation. Subsequently, a number of randomized trials of high-frequency ventilation have been conducted but have not settled the concerns raised by the HIFI study.^3,4 However, when excluding the HIFI trial (which was not using a recruitment strategy) from the meta-analysis, high-frequency ventilation has not shown to increase the risk of ICH.³ When looking only at RCTs using the identical frequency oscillatory ventilator as we used (ie, Sensor Medics 3100A, Sensor Medics Inc, Anaheim, CA), versus conventional ventilation (6 studies, 591 patients), the odds ratio for ICH for a premature strata, after stratification for the effects of maturity, was 0.79 (confidence limit: 0.59-1.08) in favor of high-frequency oscillation HFO.⁵ Furthermore, we cannot find any evidence in the published literature supporting the hypothesis that a volume recruitment strategy (the authors of the letter call this an "overly aggressive HFO" strategy) does increase the risk for ICH. However, we agree that larger trials are required to clarify this concern.

The statement regarding an increased risk for airleaks is based on the experience of a single group.⁶ It must be well-recognized that this group of investigators (Thome and Pohlandt) used a different high-frequency ventilation device (ie, high-frequency flow interrupter) than we did (ie, high-frequency oscillator). The latter uses a fully active expiration whereas the former has the characteristics of a flow interrupter (passive expiration properties) combined with a Venturi system to generate a negative pressure swing during the expiratory phase. Although the same authors showed in previous work that such a ventilator design allows the avoidance of gas trapping and lung hyperinflation,^7,8 they fail to explain why they did observe in their RCT⁶ such an increased incidence of air leak in both groups. In this regard our historical group performed even better despite a relative injurious ventilation mode. However, it has to be kept in mind that it is difficult to compare different patient groups from different centers, ventilated with different types of ventilators, frequencies, and inspiratory and expiratory times.

Concerning the incidence of chronic lung disease (CLD) in our trial, the authors of the letter make many assumptions. It is hard, even somewhat very artificial and nonscientific, to make an argument with a withdrawal rate of 5 versus 4 patients, corresponding to calculated percentages of 16% versus 10%, in favor for the HFO group regarding outcome (ie, CLD). The numbers are simply too small. To avoid this problem, we calculated the maximal possible frequency of CLD in our HFO group to be at 0% to 11% (95% CI). Furthermore, by postulating that all infants in whom therapy was withdrawn early would have developed CLD, the maximal incidence of CLD in the HFO group would have reached only 15%. Even this artificial number would remain below the actual numbers quoted in the literature.

We did not imply in our discussion section that Thome et al⁶ crossed-over 50% of their patients. We only discussed the fact that they observed many treatment failures and that this may have influenced the outcome data because a cross-over was allowed, especially when the analysis involves an intention-to-treat approach without separating the patients who were changed to the alternative ventilator strategy. The authors disclose now that they crossed-over 13% of their patients, which I believe would have been important to publish and analyze separately in their RCT⁵ to avoid any possible bias.

One of the intentions of our presentation was to show how important it is to specifically review the results of a change in a treatment strategy, which was for us a logical reaction to the results of the Provo trial.⁹ We can confirm that from September 1998 until June 2000 we did not observe a single patient with a CLD after being ventilated with HFO, having almost doubled our patient numbers in the meantime in the same patient group.

Our observational data will need confirmation by well-conducted trials with larger patient numbers. What we will certainly need is an appropriate concept of an early lung protective strategy during conventional ventilation, using a concept of opening up the lung and keeping it open, according our own experimental data^10-12 and the recent promising experience in adult acute respiratory distress syndrome patients.¹³

Peter Rimensberger, MD
Pediatric and Neonatal Intensive Care Unit
Hopital des Enfants
University Hospital of Geneva
CH-1211 Geneva 14, Switzerland

REFERENCES

1. Rimensberger PC, Beghetti M, Hanquinet S, Berner M First intention high-frequency oscillation with early lung volume optimization improves pulmonary outcome in very low birth weight infants with respiratory distress syndrome. Pediatrics. 2000; 105:1202-1208
2. HIFI Study Group High-frequency oscillatory ventilation compared with mechanical ventilation in the treatment of respiratory failure in preterm infants. N Engl J Med. 1989; 320:88-93
3. Clark RH, Dykes FD, Bachman TE, Intraventriucular hemorrhage and high-frequency ventilation: a meta-analysis of prospective clinical trials. Pediatrics. 1996; 98:1085-1061
4. Clark RH, Gerstmann DR Controversies in high frequency ventilation [review]. Clin Perinatol. 1998; 25:113 [Medline]
5. Null DM, Bachmann TE, Ashurst JT. A meta-analysis of the outcomes of randomized controlled trials of the 3100 A high frequency oscillatory ventilator. Third European Conference on High Frequency Ventilation; October 20-22, 1999; Ovifat, Belgium. Abstract
6. Thome U, Kössel H, Lipowsky G, Randomized comparison of high frequency ventilation with high rate intermittent positive pressure ventilation in preterm infants with respiratory failure. J Pediatr. 1999; 135:39-46
7. Thome U, Töpfer A, Schaller P, Pohlandt F Effect of mean airway pressure on lung volume during high frequency ventilation in preterm infants. Am J Respir Crit Care Med. 1998; 157:1213-1218 [Abstract/Free Full Text]
8. Thome U, Pholandt F Effect of TI/TE ratio on mean intratracheal pressure in high frequency oscillatory ventilation. J Appl Physiol. 1998; 84:1520-1527 [Abstract/Free Full Text]
9. Gerstmann DR, Minton SD, Stoddard RA, The Provo multicentre early high frequency oscillatory ventilation trial: improved pulmonary and clinical outcome in respiratory distress syndrome. Pediatrics. 1996; 98:1044-1057
10. Rimensberger PC, Cox PN, Frndova H, Bryan AC The open lung during small tidal volume ventilation: Concepts of recruitment and "optimal" PEEP. Crit Care Med. 1999; 27:1946-1952 [CrossRef][Medline]
11. Rimensberger PC, Pristine G, Mullen JBM, Cox PN, Slutsky AS Lung recruitment during small tidal volume ventilation allows minimal PEEP without augmenting lung injury. Crit Care Med. 1999; 27:1940-1945 [CrossRef][Medline]
12. Rimensberger PC, Pache JC, McKerlie C, Frndova H, Cox PN Lung recruitment and lung volume maintenance: a strategy for improving oxygenation and preventing lung injury during both conventional mechanical ventilation and high-frequency oscillation. Intensive Care Med. 2000; 26:745-755
13. The Acute Respiratory Distress Syndrome Network Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med. 2000; 342:1301-1308 [Abstract/Free Full Text]