Dear Sirs,
We believe we have addressed all concerns raised by the reviewers, and incorporated them
into the revised manuscript. Please find below a point by point reply to the issues raised by
the reviewers.
Sincerely,
Dr. Jan Hendrickx and co-authors.
Reviewer's report
Title: Can Modern Infrared Analyzers Replace Gas Chromatography to Measure
Inhaled Anesthetic Vapors Concentrations?
Version: 1 Date: 24 September 2007
Reviewer: Girish Joshi
Reviewer's report:
General
The authors compare the newer IR monitors for inhaled anesthetics with the 'gold' standard
gas chromatography. This is a well-written manuscript. The authors have provided some
discussion regarding the clinical relevance of their findings, such as potential overdosing at
lower concentrations. However, it would be beneficial for anesthesiology practitioners to
know clearly what they should do if faced with a IR monitor in the operating room. With
increasing interest in 'awareness' during general anesthesia, and attention towards gas
monitoring as well as brain function monitoring, the authors could comment on the potential
implications of their findings on the possibility of awareness if the anesthesia practitioner
were to use gas monitoring to guide 'depth of hypnosis.'
We believe this issue is addressed in the Discussion: “At low concentrations however, in
the MACawake range, one is more concerned about the potential for recall, especially in
patients who cannot sustain haemodynamic stability in the presence of anesthetic
vapors. A measurement that is lower than the actual concentration may lead the
clinician to increase the concentration unnecessarily and/or use a vasopressor to
maintain adequate blood pressure for the given depth of anesthesia. An error on the
high side may lead to inadequate vapor delivery to prevent recall.”
-------------------------------------------------------------------------------
Reviewer's report
Title: Can Modern Infrared Analyzers Replace Gas Chromatography to Measure
Inhaled Anesthetic Vapors Concentrations?
Version: 1 Date: 7 October 2007
Reviewer: Philip Peyton
Reviewer's report:
General
------------------------------------------------------------------------------Major Compulsory Revisions (that the author must respond to before a decision on
publication can be reached)
1. Documentation of reliability and precision of GC (and IR):
The paper seems to report the scatter in agreement of individual measurements by GC and IR
across a range of gas concentrations and species. My major concern with this study is the
authors' assumptions regarding the use of GC as a standard for the assessing the accuracy of
IR. The paper seems to largely ignore the issue of reproducibility (precision) of either GC or
IR, which makes interpretation of the results difficult. For example, the authors state that the
standard deviation (SD) of repeat GC measurements is said to be 2-3% by Dr. Eger.
However, they don't document the precision of their own GC measurements, which would be
a simple thing to do, taking multiple samples from a given preparation at each concentration.
We used the actual equipment (including calibration gas and gas chromatograph) and
technique of Dr. Eger. For each agent, a different cylinder with calibration gas is
prepared (secondary standard). The exact concentration of this secondary standard is
determined by comparing it with a primary volumetric standard using gas
chromatography. The primary standard is prepared by injecting a known exact amount
of liquid agent (measured by weight or glass micropipette) in a glass Erlenmeyer of
known volume, taking the effect of temperature into account. Using this primary
standard, it was determined that the concentrations of isoflurane, sevoflurane, and
desflurane in the cylinders were 0.744, 1.49, and 4.20%, respectively. Before an after
testing each agent with each analyzer, gas from the cylinder with the appropriate
calibration gas was injected 3 times to ensure consistency/precision. Based on 6
calibrations per agent per analyzer (3 before and 3 after measuring each agent) a total
of 18 GC calibrations were done per agent with the same calibration gas. Over the
entire course of the study, the peak height with the same calibration gas for isoflurane
(0.744%), sevoflurane (1.49%), and desflurane (4.20%) was 54.0 mm (standard
deviation 1.1 mm) for isoflurane, 52.2 mm (standard deviation 2.0 mm) for sevoflurane,
and 54.1 mm (standard deviation 1.4mm) for desflurane. The coefficient of variation for
isoflurane, sevoflurane, and desflurane therefore was 2.0, 3.8, and 2.6 % respectively.
This information has been added to the Methods section.
There appears to be considerable scatter in GC measurements along the x-axis in
Figure 1.
The variation the reviewer refers to is due to the fact that the exact concentration of the
vapor at the common gas outlet slightly differed for each analyzer. For each analyzer,
test gases were sampled from the common gas outlet at different times; because
vaporizer output (and thus the concentration of vapor at the common gas outlet) may
differ slightly from moment to moment, some scatter along the X-axis from figure 1 will
be due to the manner in which the test gas was prepared. This information has been
added to the Methods section.
If the SD of GC = 3%, so 2XSD = 6% either side of the true value. Therefore up to 6% of the
scatter in agreement with IR may arise from the GC standard itself.
For GC, the coefficient of variation for isoflurane, sevoflurane, and desflurane was 2.0,
3.8, and 2.6 % respectively – see above.
Where differences between IR and GC of around 10-20% are being found, should not
multiple measurements from each sample be made at each concentration to narrow these GC
confidence limits?
Of each mixture, we had two samples analyzed by IR and two by GC.
At each concentration, how many repeat measures were made (if any)?
Over the entire course of the study, repeat measures were made on three occasions
because operator error had caused gross errors. This information has been added to the
manuscript. On 8 occasions, repeat testing was done because the sampling rate of the IR
(200 mL.min-1) was so fast relative to the volume of the syringe that the syringe had run
empty and no sample was left for the second GC determination.
What are the confidence limits for a given measurement by either GC or IR? They, quite
rightly, used the first and last 10 mL of each gas sample for GC analysis and rejected the
measurement if these differed by more than 1mm. Perhaps they can tell us how much
proportional scatter in GC measurements 1mm represents.
Over the entire course of the study, the peak height with the same calibration gas for
isoflurane (0.744%), sevoflurane (1.49%), and desflurane (4.2%) was
- isoflurane 54.0 mm (standard deviation 1.1)
- sevoflurane 52.2 mm (standard deviation 2.0)
- desflurane 54.1 mm (standard deviation 1.4%)
One mm therefore presents 1.9, 1.9 and 1.8% proportional scatter in GC for isoflurane,
sevoflurane, and desflurane measurements. These calculations are based on 6
calibrations, 3 before and 3 after measuring each agent for each analyzer (total of at 18
per analyzer).
2. Details of IR sampling:
How was data obtained from the IR device? Was it simply read from the screen of the
analyzer? How many digits of precision (at the low concentrations particularly) does this
provide?
It was downloaded in a spreadsheet every 10 seconds, so 2 or 3 values are obtained per
tested sample. The number of digits of precision is 3.
How stable was the measured concentration of gas sampled from the glass syringe during this
process?
To ensure homogeneous mixing, the content of a 50 mL glass syringe was mixed by
repeatedly ejecting and aspirating its content into another 50 mL glass syringe using a
three-way stopcock. This is mentioned in the Methods: “To ensure adequate mixing of
the gases, the mixtures were injected via a three-way stopcock at least 4 times into a
second 100 mL glass syringe.” The Methods section also mentions that the difference
between the GC before and after the IR had to be less than 1 mm. This gave us
additional certainty that the sample was adequately mixed.
How did they deal with the problem of the negative pressure generated within the syringe
during sampling (this would occur rapidly, since the sampling rate of the IR device is about
200 mL/min). This would potentially cause significant artefact. Inevitable sticking of the
plunger of the syringe during sampling would lead to considerable variation in measured
partial pressure, and would not be fair test of the device.
Both during storage and sampling, the 50 ml glass syringe was placed with the plunger
upwards. The weight of the plunger, combined with the very low resistance of the
plunger/syringe interface as well as the application of a rotating motion (by hand) on
the plunger while sampling ensured that the plunger never got stuck. The syringes did
not stick, and negative pressure did never develop.
Did they consider using analog download of their IR concentration data to computer memory
for inspection and analysis of their concentration waveforms?
No. Dr. Peyton himself only recently described this approach (Peyton PJ, Chong M,
Stuart-Andrews C, Robinson GJ, Pierce R, Thompson BR. Measurement of anesthetics
in blood using a conventional infrared clinical gas analyzer. Anesth Analg
2007;105:680-7). Our paper is in that regard complementary to his findings – IR can be
used for research purposes if the individual IR analyzer is calibrated against a
volumetric standard. The Discussion does mention that “An alternative interpretation
of Peyton’s findings might be that IR analysis could be used if the performance
characteristics of the individual IR analyzer are well documented”.
------------------------------------------------------------------------------Minor Essential Revisions (such as missing labels on figures, or the wrong use of
a term, which the author can be trusted to correct)
Methods:
The authors say that they calibrated the GC against secondary (cylinder) standards which had
been themselves been calibrated against a primary volumetric standard. I accept this although
the details of the preparation of the primary standard are not given.
Grammatical corrections:
Title: Change to “… Anesthetic Vapor Concentrations?” or “concentration of anesthetic
vapors?”.
The title has been changed accordingly.
P10. Last paragraph. “clinically” not “clinical”.
The text has been changed accordingly.
Figures:
Fig 1 “Company limits”? Explain this term in the Figure legends as well as the text, please.
We believe that this was already mentioned in the Discussion: “The performance of the
four modules conformed to that specified by the company, +/- (0.15% + 5% of IR
reading). We also added it to the legend of figure 1.”
Reviewer's report
Title: Can Modern Infrared Analyzers Replace Gas Chromatography to Measure
Inhaled Anesthetic Vapors Concentrations?
Version: 1 Date: 20 September 2007
Reviewer: Joseph F Antognini
Reviewer's report:
General
------------------------------------------------------------------------------Major Compulsory Revisions (that the author must respond to before a decision
on publication can be reached)
Page 6, near bottom: Provide more details about how the samples were prepared. Where were
the samples drawn from the machine?
The gas samples were drawn from the common gas outlet – this information has been
added to the manuscript.
Did you flush the syringe several times before obtaining the sample?
Yes.
What was the make and model of the anesthesia machine? The vaporizers?
The ADU Anesthesia Delivery Unit by GE, Helsinki, Finland; the vaporizer is an
integral part of the ADU. This information has been added to the Methods section.
Page 7, near top: You state that > 1 mm variation in peak height resulted in
repeating testing, but how much is 1 mm compared to the overall peak? Is 1 mm
a little or a lot?
How much “1mm” means, does indeed depend on the peak height. The concentration of
isoflurane, sevoflurane, and desflurane in the tank containing our secondary standard
was 0.744, 1.49, and 4.20%, respectively. Over the entire course of the study, the peak
height with the same calibration gas for isoflurane (0.744%), sevoflurane (1.49%), and
desflurane (4.20%) was 54.0 mm (standard deviation 1.1) for isoflurane, 52.2 mm
(standard deviation 2.0) for sevoflurane, and 54.1 mm (standard deviation 1.4%) for
desflurane. One mm therefore presents 1.9, 1.9 and 1.8% proportional scatter in GC for
isoflurane, sevoflurane, and desflurane measurements. These calculations are based on
6 calibrations, 3 before and 3 after measuring each agent for each analyzer (total of at
18 per analyzer). This information has been added to the manuscript.
How often did you need to repeat testing?
Over the entire course of the study, repeat measures were done on three occasions
because operator error had caused gross errors. This information has been added to the
manuscript. On 8 occasions, repeat testing was done because the sampling rate of the IR
(200 mL.min-1) was so fast relative to the volume of the syringe that the syringe had run
empty and no sample was left for the second GC determination.
Provide more details about the calibration of the GC. What was the range of concentrations of
the calibration gases? How exactly were these prepared?
The same calibration gas (one for each agent) was used throughout the study. The
calibration gas comes out of a stainless steel cylinder which contains a known
concentration of vapor; for each agent, a different cylinder was prepared. The mixture
is prepared by creating a subatmospheric pressure in the cylinder. Next, an amount of
liquid agent is aspirated into the cylinder. This amount is calculated such that after
vaporization the calculated pressure in the cylinder will remain below the vapor
pressure of the agent. The cylinder is then filled/pressurized with air. To mix its
contents, the cylinder is rolled several times, and subsequently heated overnight by
applying an external heat pad. The exact concentration of this secondary standard is
determined by comparing it with a primary volumetric standard using gas
chromatography. The primary standard is prepared by injecting a known exact amount
of liquid agent (measured by weight or glass micropipette) in a glass Erlenmeyer of
known volume, taking the effect of temperature into account. Using this primary
standard, it was determined that the concentrations of isoflurane, sevoflurane, and
desflurane in the tank were 0.744, 1.49, and 4.20%, respectively. This information has
been added to the manuscript.
When were the modules last serviced relative to when you tested them? Is the variation due to
some modules being recently serviced, while others were not?
Drift over time was not examined, and therefore cannot be addressed.
I am not sure what statistical analysis you did to make the claim in table 1 that individual
modules differed. Yes they differed, but what test did you use to come up with p=0.004?
We believe this was addressed in the Methods section: “To examine the effect of carrier
gas and whether analyzers differed in their performance, we used two way ANOVA
(factors = carrier gas and analyzer) with the Holm-Sidak method”.
When looking at the GC analysis, there is the same amount of variation. So, isn’t it possible
there is a problem with there being variation in the way the samples were injected? Please
clarify.
We don’t think so - the variation the reviewer refers to is due to the fact that the exact
concentration of the vapor at the common gas outlet slightly differed for each analyzer.
For each analyzer, test gases were sampled from the common gas outlet at different
times; because vaporizer output (and thus the concentration of vapor at the common
gas outlet) may differ slightly from moment to moment, some scatter along the X-axis
from figure 1 will be due to the manner in which the test gas was prepared. This
information has been added to the Methods section.
-------------------------------------------------------------------------------