Statistics from Altmetric.com
I was surprised to find that the paper by Joneset al 1 did not come up to the normally high standard of the journal. Firstly, the study design itself lacks a control group and yet attempts to make a contribution to the debate about using non-invasive positive pressure ventilation (NIPPV) to improve survival in COPD. Clearly, comparing before and after treatment as a measure of survival can be misleading as it may introduce a “placebo effect” bias that may easily swamp out any physiological improvement. In these days of systematic reviews this evidence is flimsy and may well dilute the effect of good prospective controlled randomised studies.
There were a few minor errors in table 1 which, firstly, presents the transfer factor (Tlco) in non-SI units and offers no conversion factor as quoted in the instructions for authors. Secondly, the units for the transfer coefficient (Kco) are expressed in ml/min instead of mmol/min/kPa/l. I would also have preferred to see the lung volumes, that were expressed as percentage of predicted, have their measured units as litres instead of just “percentage” of predicted value.
Despite these shortcomings, I strongly support the final paragraph that more prospective studies on larger patient groups with clearly defined enrolment criteria are desperately needed to help secure essential funding to what providers of this service all believe is effective medical care.
authors’ reply We thank Dr Cooper for his interest in our paper and agree that a control group might have strengthened our conclusions. However, to establish such a group with a sham device for NIPPV over the time periods involved would, in practice, be difficult. We are aware of only one study1-1 in which a sham device was used. No improvements in arterial blood gas tensions were found over a three month follow up period but levels of pressure support were low in the treatment group with inspiratory pressures of 10 cm H2O, there was not sufficient monitoring to allow correction of CO2, and the starting Paco 2 was only mildly raised to a mean of 48.5 mm Hg. In our group of patients who had more severe COPD, inspiratory pressures of up to 20 cm H2O were required and our mean status Paco 2 was 8.0 kPa. A technical intervention such as NIPPV might result in a reduction in hospital or general practitioner visits through a “placebo effect” (conversely, dependency on technology might have resulted in more attendances due to anxiety or technical failings). However, we do not accept that physiological measurements such as improved arterial blood gas tensions or worsening lung function are affected by placebo.
In an attempt to obtain a control group we compared our patients with those from the MRC and NOTT trials; these well recognised studies of LTOT investigated a population similar to our own in terms of baseline characteristics, as stated in our discussion. Without the option of NIPPV, our patients would have remained on LTOT which would have been insufficient to correct hypoxia.
The SI units for transfer factor are mmol/min/kPa; the conversion factor is to multiply by 0.335 or divide by 2.986. Transfer coefficient units are mmol/min/kPa/l. The lung volumes in table 1 are quoted in both litres and percentage predicted.
Finally, as Dr Cooper states, most of us working in this area who see the improvement NIPPV makes to patients with severe hypercapnic respiratory failure due to COPD believe that it is an effective medical treatment. Multicentre European trials comparing LTOT with LTOT and NIPPV are currently underway and the results should help to clarify the role of domiciliary NIPPV in COPD.