eLetters to:

ARTICLE: Peter C. Rimensberger, Maurice Beghetti, Silviane Hanquinet, and Michel Berner 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] [Full text] [PDF]

eLetters: Submit a response to this article

eLetters published:

Weak data leads to wrong message

Ulrich Thome   (9 June 2000)

Response to: Weak data leads to wrong message

Peter C Rimensberger   (20 June 2000)

Weak data leads to wrong message 9 June 2000
Ulrich Thome, Neonatology Fellow University of Alabama at Birmingham Send letter to journal: Re: Weak data leads to wrong message uthome{at}uab.edu Ulrich Thome	Dear Dr. Lucey: We were surprised of finding an article of high frequency oscillation (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 a study, PEDIATRICS sends a false message based on weak data to the caretakers of premature infants. The danger of overly aggressive high frequency oscillation (HFO) lies in an increased incidence of air leaks (2) 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%), not significant due to the small sample size, but worrisome in light of the de-clining incidence of ICH worldwide. The reduction of chronic lung disease (CLD) in the study of Rimensberger et al. (1) may be ex-plained by three 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 al. (1) suggest that the outcome in their HFO group was im-proved because of limited exposure of the infants to conventional ventilation. However, their pa-tients 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 group (1) in comparison with the HFO treated infants in our recent randomized trial (2), which are unfortunately not discussed by the authors, are the fre-quent withdrawal of support, thus preventing the survival of the highest risk infants, and higher birthweight and gestational age (mean gestational age 27.7 weeks vs. 26.9 weeks, mean birth-weight 981g vs 913g). Also, reporting only mean ± SD does not tell us whether infants with a gestational age below 25 weeks, i.e. those with the highest risk for developing CLD, were included. There were 36 such infants in our study (2). On page 1207, Rimensberger et al. (1) falsely imply that 50% of our patients (2) 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. Sincerely Dr. Ulrich Thome Division of Neonatology Department of Pediatrics University of Alabama at Birmingham 525 NHB, 619 S. 19th St. Birmingham AL 35233-7335 Prof Dr. F. Pohlandt Division of Neonatology University Children's Hospital 89070 Ulm Germany Reference List 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. Thome U, Kössel H, Lipowsky G, et al. 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. 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. HiFO Study Group. Randomized study of high-frequency oscillatory ventilation in infants with severe respi-ratory distress syndrome. J Pediatr. 1993;122:609-619. Moriette G, Walti H, Salanave B, et al. Prospective randomized multicenter comparison of high-frequency oscillatory ventilation and conventional ventilation in preterm infants <30 weeks gestational age with RDS. Pediatr Res. 1999;45:212A(Abstract)
Response to: Weak data leads to wrong message 20 June 2000
Peter C Rimensberger, Pediatric and Neonatal ICU University of Geneva, Switzerland Send letter to journal: Re: Response to: Weak data leads to wrong message peter.rimensberger{at}hcuge.ch Peter C Rimensberger	In response to the letter intitled « weak data leads to wrong message » from Thome and Pohlandt we would like to take position and discuss some of the concerns adressed 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 assure 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 discussion on the risk of intracranial hemorrhages (ICH) started with the multicenter HIFI study (2) 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 high frequency oscillation 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 (CL 0.59 - 1.08) in favor of high frequency oscillation (5). Furthermore we can not find any evidence in the published literature supporting the hypothesis that a volume recruitment strategy (the authors 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 the group by Thome and Pohlandt used a different high frequency ventilation device (i.e. high frequency flow interrupter) than we did (high-frequency oscillator). The latter uses a fully active expiration whereas the fromer 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 to avoid gas trapping and lung hyperinflation (7, 8) they fail to explain why they did observe in their RCT (6) 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 in our trial, the authors of the letter make many assumptions. It is hard, even somewhat very artifical and non scientific, to make an argument with a withdrawal rate of 5 vs 4 patients, corresponding to calculated percentages of 16 vs 10%, in favor for the HFO group regarding outcome (i.e. CLD). The numbers are simply to 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). Furhtermore, if we would postulate that all babies in whom therapy was withdrawn early, would have developped CLD, the incidence of CLD in the HFO group, in the worst case, would have reached only 15%. Even these artifical numbers would remain below the actual numbers quoted in the literature. We did not imply in our discussion section that Thome et al (6) did crossover 50% of their patients. We only discussed the fact, that they observed many treatment failures and that this may have influence the outcome data because a crossover was allowed, especially when the analysis are made on an intention to treat approach without separating the patients who were changed to the alternative ventilator strategy. The authors tell us now that they crossed over 13% of their patients, which I believe would have been important to publish and analyse separately in their RCT (5) 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 agree that our results are impressive and hard to believe, but we can confirm that from September 1998 until today (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. These observational data will need confirmation by well conducted trials in larger patient numbers. What we will certainly need is an appropriate concept of an early lung protective strategy during conventional ventilation. Peter Rimensberger, MD Pediatric and Neonatal ICU Hopital des Enfants University of Geneva CH 1211-Geneva, 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 infnats with respiratroy 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, et al. 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 5) Null DM, Bachmann TE. A meta-analysis of the outcomes of randomized controlled trials of the 3100 A high frequency oscillatory ventilator. (Abstract). Third European Conference on High Frequency Ventilation, October 20 - 22, 1999, Ovifat, Belgium 6) Thome U, Kössel H, Lipowsky G et al. 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 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 9) Gerstmann DR, Minton SD, Stoddard RA, et al. The Provo multicentre early high frequency oscillatory ventilation trial: improved pulmonary and clinical outcome in respiratory distress sydnrome. Pediatrics 1996; 98:1044-1057