EDITOR:
The paper by Cannesson and colleagues [Reference Cannesson, Desebbe, Hachemi, Jacques, Bastien and Lehot1], comparing respiratory-associated variations in the optical plethysmograph and arterial pulse pressure, was very interesting. It was a tightly controlled study. A study I report here, carried out in New Zealand and in the UK, was of a more uncontrolled clinical nature [2].
With the agreement of local research Ethics Committees in Auckland (NZ) and Cambridge (Addenbrooke’s, UK) and the patients’ consent, electronic data were collected from anaesthetic monitors (Datex-Ohmeda Division, Instrumentarium Corp., Helsinki, Finland) that were used during major surgical procedures. The scenarios were true clinical situations without standardization of anaesthetic technique. One set of data was collected at 10-s intervals to give an overall physiological picture of events, another set was collected at 100 Hz to enable waveforms of arterial pressure and pulse-oximeter plethysmography to be analysed and free text comments were collected from the anaesthetist. The data were stored on a disk and analysed off-line.
The data were smoothed to enable timestamping of the systolic peaks and then the original values retrieved using this timestamp. Long-duration artefacts, such as arterial blood sampling, were removed manually. The respiratory variability of the arterial pressure was compared with the variability of the plethysmograph trace in several ways. Twenty-second epochs (enclosing at least three respiratory cycles) were used to determine the periodic variability. An adjustment to the raw wave-form data was made to compensate for any general upward or downward trend in the analysis period.
Systolic pressure variation and pulse pressure variation were both investigated. The plethysmogram, not having a zero baseline, was analysed by determining the envelopes that characterized the peaks and troughs, and in another analysis only the peak values were used.
No matter which technique was used, the correlations between pressure variation and plethysmographic variation were very poor and the Bland–Altman plots indicating that the substitution of respiratory related plethysmographic variation for pressure variation are inappropriate.
The graph (see Fig. 1) represents 90 min of one patient’s data (nine 10-min periods during a 5-h procedure). Each measurement is from a 20-s epoch.
Bland–Altman plot of respiratory-related plethysmographic variation (%) and respiratory-related systolic pressure variation (%).
The results above are typical and in my view it made the use of the plethysmographic waveform variability not an appropriate substitute for respiratory-associated arterial pressure variation, I believe their comments about the drawbacks of arterial pressure monitoring are also overstated. This study was not as well-controlled as that by Cannesson and colleagues, but was carried out in realistic clinical situations. I think the clinical usefulness of respiratory-associated changes in the optical plethysmograph is still to be proven.
