Article Text
PDF
Statistics from Altmetric.com
Editor—Baumer reports the results of a large multicentre study comparing the effects of patient triggered ventilation (PTV) with conventional ventilation (IMV).1There appears to be no benefit from PTV compared with IMV in death rate, development of chronic lung disease, pneumothorax rates, and cerebral ultrasound abnormality. In addition, because of an increased trend toward a higher pneumothorax rate, Baumer concludes that, at present, PTV delivered with either the SLE 2000 or the Dräger babylog 8000 ventilators cannot be recommended for infants of less than 28 weeks gestation with respiratory distress syndrome (RDS).
However, we are concerned that this may be a premature conclusion given the significant difference in PTV delivered by the two main ventilators used and the potential heterogeneity of clinical practice within the different centres involved, despite agreed ventilation protocols. Dimitriou et al 2 showed that neonates and infants trigger a significantly lower proportion of breaths using the SLE 2000, an airway pressure triggered ventilator, compared with the Dräger babylog 8000, an airflow triggered ventilator that provides synchronised intermittent positive pressure ventilation (SIPPV). Attempts to optimise the trigger rate of the SLE 2000 ventilator by increasing pressure sensitivity often results in autotriggering as discussed by Baumer. Therefore the PTV modes of the two ventilators are substantially different. This prompts us to ask whether the findings of this multicentre study are only applicable to PTV provided by the SLE 2000? Would there have been a different outcome if all the triggered babies had received SIPPV?
It is not known how many of the 40/213 babies of less than 28 weeks gestation who had pneumothoraces were receiving SIPPV. As only 11% (52/465) of all triggered babies ever received SIPPV, we surmise that very few of the 40 were ventilated in this way. Is it fair to conclude that the Dräger babylog 8000 has a trend to pneumothorax on SIPPV?
In a separate smaller study, Baumer also reports 120 patients in three centres randomly assigned to either the Dräger babylog 8000 or the SLE 2000 ventilator, and found a non-significant trend to higher pneumothorax rate, chronic lung disease, and death for the former group. But we are not told how many of the babies were actually on trigger mode (PTV or SIPPV). They could all have been receiving IMV on the Dräger babylog 8000. Therefore is it possible that SIPPV is not being tested?
A further finding was a significantly higher rate (124/463) of triggered babies that departed from their assigned mode of ventilation, 45 of these failing to trigger their ventilator. Were they all on the SLE 2000 ventilator, as Dimitriou et al 2 would predict?
Finally, we note that 10 of the 22 neonatal units each recruited less than 20 patients over the four year period, one contributing only one patient. Could the technique of PTV ventilation in units contributing so few be different from those enrolling 60–136 neonates over the same period, despite prior visits from the trial coordinator? Would a logistic regression for morbidity against number of patients contributed from each unit show that the greatest morbidity occurred in units that contributed fewer patients, rather than those using PTV or SIPPV modes?
Given the heterogeneity of the units involved and the significant difference in ventilators used, we think that it is premature to dismiss SIPPV on the Dräger babylog 8000 in neonates less than 28 weeks gestation with RDS. We agree with Baumer that further studies are required, and extend his conclusion by saying that PTV with the SLE 2000 (n = 411) rather than SIPPV from the Dräger babylog 8000 ventilator (n = 52) cannot be recommended in this group.
Dr Baumer responds on behalf of the trial collaborators: The interest displayed in the trigger ventilation trial by Burmester and Petros is welcome. Their letter raises questions about the interpretation of the performance of the Dräger babylog 8000, which was used in a minority of infants in the study.
The trigger sensor device is different, and I agree with their implied statement that, as each trigger ventilator performs differently, results obtained using one ventilator cannot be extrapolated to another. However, both ventilators were used in PTV mode (sometimes referred to as synchronised intermittent positive pressure ventilation, assist control, or synchronised assisted ventilation in infants), with the ventilator set to trigger at each inspiratory effort. No infants in this study were ventilated with SIMV (synchronised intermittent mandatory ventilation) where the baby's breaths, selected during a “time window”, trigger the ventilator with the preset number of breaths per minute.
The results were reported in the way that they were for a reason. The original study design allowed a four way randomisation between the two makes of ventilator and the two modes of ventilation. As was reported, only three centres had enough of both ventilators to allow this to occur. Other centres ventilating infants with the Dräger babylog had a two way randomisation between PTV and IMV.
Therefore the possibility of confounding by differences in practice between centres needed to be excluded. If the centres using the Dräger as well as the SLE ventilator had different outcomes from those using only one make of ventilator, this may have led to inappropriate conclusions being drawn if all infants being ventilated with one ventilator were simply compared with those being ventilated with another. A logistic regression model was therefore used to allow for possible centre effects (as well as other significant factors such as gestation).
With that caveat, I have extracted the numbers from the database giving details of the crude observed rates of pneumothorax in the infants of less than 28 weeks gestation, separately reported for the two makes of ventilator (table 1-1).
Rates of pneumothorax in infants of less than 28 weeks gestation using two different makes of ventilator
The observed rate of pneumothorax was substantially (but not significantly, χ2 2.8, p > 0.05 < 0.1) higher in the infants ventilated in PTV mode than in IMV mode using the SLE 2000. Although the numbers are small, the observed pneumothorax rate was higher in infants ventilated with the Dräger babylog 8000 than in those ventilated with the SLE ventilator. It therefore seems somewhat illogical to recommend caution in using the SLE 2000 in PTV mode in infants less than 28 weeks gestation, but not extend this caution to ventilating infants with the Dräger babylog 8000.
Given that none of these differences were statistically significant, no clear recommendation can be given. This is why the wording used in the publication was “it might be prudent to avoid . . .”.
As regards the number of infants departing from their assigned mode of ventilation, several points should be emphasised. The study protocol permitted changing the mode of ventilation at the discretion of the attending clinicians. This was inevitably interpreted differently by each clinical team. Departure from the assigned mode of ventilation was not an intended outcome, and it is evident that this occurred more commonly in the more immature infants and those that subsequently died. High rates of departure from the assigned mode cannot therefore readily be interpreted as evidence of failure of the assigned mode. Table 1-2shows the numbers of infants of all gestations departing from the assigned mode of ventilation.
Numbers of infants of all gestations departing from the assigned mode of ventilation
There was therefore a higher crude rate of departure from the assigned mode of ventilation in infants ventilated with the Dräger babylog 8000, with a similar proportion transferred for failure to trigger.
It would be difficult to interpret the pneumothorax rates for infants who were actually being ventilated with their assigned mode of ventilation. Some infants were switched to another mode of ventilation after sustaining their pneumothorax. Most of the pneumothoraces occurred while infants were receiving their assigned mode of ventilation, and this included infants being trigger ventilated using the Dräger ventilator.
Burmester and Petros ask whether centres contributing few patients may have higher morbidity rates, correcting for potential confounding factors by using a logistic regression model. The pneumothorax rate from centres contributing less than 20 patients was the same as the centres contributing more infants.
We have used a model to identify outcome differences in infants randomised within three months of the first infant being entered into the study, correcting for individual centre effects, gestation, birth weight, and mode of ventilation. There was no significant difference in rates of death and chronic lung disease, abnormal cranial ultrasound scan, or duration of ventilation. However, an appreciable and statistically significant difference was found for pneumothorax rates. The 139 infants randomised within three months had a pneumothorax rate of 5% compared with a rate of 13% for those randomised more than three months into the trial (odds ratio 0.30; 95% confidence intervals 0.12 to 0.74; p = 0.009). This was seen equally for both modes of ventilation.
This finding suggests that the initial educational visit by the trial coordinator had a beneficial effect on ventilator management which disappeared as infants continued to be enrolled.
In summary, there is no evidence from this study of any trend towards better outcomes with the Dräger babylog 8000 ventilator, although the small numbers enrolled make any conclusions less robust. There is evidence that suggests there may have been a short term reduction in pneumothorax rates from the educational package offered at the start of the trial.
In conclusion, there was no convincing evidence of a beneficial effect of a policy of using PTV in preterm infants with RDS with the ventilators used. Regular attention to staff education on ventilator techniques is recommended.
I would like to use this opportunity to pay tribute to the two trial coordinators (Sue Ellis and Tom Mill), to the trial statistician (David Wright), and to the data monitoring committee (David Field and Diana Elbourne), whose details were inadvertently omitted from the final paper and without whom, together with the trial collaborators, the study would not have been possible.