RESPONSE TIME STUDIES OF A NEW, PORTABLE MASS SPECTROMETER
Paul A. Delaney, MS,1 George M. Barnas, PhD,1; 2 and Colin F. Mackenzie, MD 1; 2
Delaney PA, Barnas GM, Mackenzie CF. Response time studies of a new, portable mass spectrometer. J Clin Monit 1997; 13: 181^189
ABSTRACT. Objective. Mass spectrometers are frequently used by anesthesiologists perioperatively to monitor patients' respiratory function and levels of inhaled anesthetics. Due to size, complexity and expense, they are typically used in a time-sharing manner which degrades their performance. We assessed the accuracy of the Random Access Mass Spectrometer (RAMS, Marquette Electronics) which is small enough to be dedicated to a single patient. Methods. We compared the 10^90% rise times for O2 , CO2 , N2 O and iso£urane for the RAMS with di¡erent catheter con¢gurations to those of a MedSpect mass spectrometer (Allegheny International Medical Technology) operating under ideal conditions. For CO2 the lag of the RAMS relative to the MedSpect was also measured. Next, perioperative conditions were simulated by ventilating anesthetized dogs with a variety of inhalatory gases and ventilatory parameters, and the interchangability of the two devices was assessed. Results. When ¢tted with a catheter with minimal dead space the MedSpect had rise times of 0.11^0.12 sec while the RAMS had rise times of 0.07^0.12 sec and a delay of 0.19 sec compared to the MedSpect. The rise times and delay of the RAMS increased when using a larger catheter and water trap. Although there were statistically signi¢cant di¡erences in some values for inhaled and end-tidal gases under simulated perioperative conditions, particularly at the higher frequencies, these di¡erences were small and for most purposes not clinically signi¢cant. Conclusions. Our results demonstrate that the RAMS con¢gured for clinical conditions performs nearly as well as the MedSpect under ideal conditions. The small di¡erences between the two, con¢ned almost entirely to their end-tidal CO2 values, could be due to di¡erences in instrument calibration, by the larger sampling catheter commonly used in clinical settings, or by a combination of both factors. Therefore the RAMS is su¤ciently accurate for clinical use and would alleviate problems associated with timeshared mass spectrometers. KEY WORDS. Gas measurement, critical care, instrument comparison
INTRODUCTION
From the Departments of 1 Anesthesiology and 2 Physiology, University of Maryland School of Medicine, Baltimore, Maryland 21201. Received Oct 8, 1996. Accepted for publication Mar 14, 1997. Address correspondence to Paul A. Delaney, Anesthesiology Research Laboratories, University of Maryland School of Medicine, 10 South Pine Street, MSTF 534, Baltimore, Maryland 21201-1192. Journal of Clinical Monitoring 13: 181^189, 1997. ß 1997 Kluwer Academic Publishers. Printed in the Netherlands.
Mass spectrometers are routinely used in clinical anesthesiology perioperatively to measure O2 , N2 O, N2 , and CO2 as well as the anesthetic agents iso£urane, halothane, en£urane, and sevo£urane. Due to their size, cost, and complexity of maintenance, it has been heretofore unfeasible to devote one mass spectrometer to each patient. As a result, most mass spectrometers used in clinical anesthesiology are time-shared, a scheme wherein samples from several operating rooms and intensive care units are measured sequentially by a single mass spectrometer. Such an arrangement o¡ers substan-
182
Journal of Clinical Monitoring Vol 13 No 3 May 1997
tial advantages in terms of costs per patient [1] and maintenance. However, the numerous disadvantages of time-shared mass spectrometers include: 1) a single mass spectrometer failure a¡ects several rooms simultaneously [2, 3]; 2) each room is sampled only a few seconds during each 1^3 minute cycle [1]; and 3) the long sampling catheters that time-shared mass spectrometers require distort and delay the signals for each room [4^8]. Time-shared mass spectrometers also provide continuous monitoring during intubation and other critical times in a single room only when speci¢cally signalled to do so, leaving the device temporarily unavailable to rooms. These problems would be minimized or eliminated by using a mass spectrometer unit that is easy to maintain and calibrate, while compact and inexpensive enough to dedicate to a single patient [9, 10]. The Random Access Mass Spectrometer (RAMS, Marquette Electronics, Milwaukee WI) has recently become available for use in clinical anesthesiology and is purported to have such features. The purpose of this study was to assess the performance characteristics and accuracy of the RAMS during simulation of perioperative conditions that would be commonly encountered during clinical use. METHODS
The mass spectrometers The RAMS weighs 34 kg and measures 26.7 26.7 45.1 cm. It quanti¢ed gas concentrations by ionizing the molecules, then directing one species at a time to a single collector, sampling each species sequentially. The manufacturer's product speci¢cations state that its sampling rate increases when fewer gases are sampled. However, we found that this phenomenon was not apparent in the RAMS's analog output, which is updated every 0.045 sec (22 Hz), regardless of the number of gases measured. For this study, the RAMS was programmed to measure O2 , N2 O, N2 , CO2 , halothane, iso£urane and en£urane. Its performance was assessed using di¡erent types of sampling catheter con¢gurations (Table 1). For comparison, we selected the MedSpect mass spectrometer (Allegheny International Medical Technology), which weighs 204 kg and measures 58.4 63.5 104.1 cm [11]. The MedSpect uses the traditional method of molecular ionization followed by magnetic de£ection [12] to measure O2 , N2 O, N2 , CO2 , halothane, iso£urane and en£urane. For all measurements in this study the MedSpect used a catheter with minimal dead space (0.003 in ID 5-ft stainless-steel).
All £ow rates were veri¢ed with a high precision £owmeter (Cole-Parmer, Niles IL) prior to the data collection since the £ow rate displayed by the RAMS con¢guration software was found to be in error by as much as 100 ml minÿ1 .
Part I We ¢rst characterized the responses of the two mass spectrometers when presented with simultaneuous and instantaneous changes in the O2 , N2 O, CO2 , and iso£urane contents. The MedSpect and the RAMS independently underwent two-point calibrations according to the manufacturer's respective guidelines for each using calibration grade gases. Calibration was veri¢ed for both using dry air (79.1% N2 and 20.9% O2 ) and another calibration grade mixture of gases (0.1% for major components, 0.05% for minor components) consisting of 30.0% O2 , 62.0% N2 O, 5.0% CO2 , 1.0% halothane, 1.0% iso£urane, and 1.0% en£urane (Marquette Electronics, Milwaukee WI). All measurements were completed within one-half hour after calibration. The analog CO2 , O2 , N2 O and iso£urane outputs for each mass spectrometer were recorded via a multichannel paper chart recorder at 100 mm secÿ1 (Astro-Med MT95000-16, East Warwick RI). A system consisting of a manually operated stopcock switching from 100% nitrogen to the calibration-grade anesthetic gas mixture attached to a small bore tube was used. Both mass spectrometers sampled the gases at points inside the tube equidistant from the stopcock. After the MedSpect and RAMS had sampled N2 for several seconds, the stopcock was turned so that they sampled the anesthetic gas mixture. The times required
Table 1. Catheter con¢gurations used on the RAMS. The MedSpect used only the ¢rst con¢guration for all measurements. Sampling rates were determined by operating speci¢cations for sample inlet pressure. The AQUA-Knot 2 [14] (Marquette Electronics, Jupiter FL) selectively removes water from gas samples passing through it Length (in)
Inside diameter (in)
Sampling rate (ml minÿ1 )
60
0.003
60
120
0.063
240
120
0.063
240
Material
AQUAKnot 2
Te£on-coated stainless steel Polyvinyl chloride Polyvinyl chloride
No No Yes
Delaney et al: ResponseTime Studies of a New, Portable Mass Spectrometer
for each signal to change from 10% to 90% of its ¢nal value (the ``rise times'') were determined from the chart recordings; this is the most typically used measure of signal distortion in mass spectrometers [10]. However, because the RAMS's output signal is the product of a digital-to-analog converter, it produces stepwise responses every 0.045 sec. For the RAMS, the 10^90% rise times were calculated from the points at which the signal ¢rst reached or exceeded these points. Each value of the rise times was taken as the average of three identical measurements. In addition, the delay of the RAMS relative to the MedSpect was determined by measuring the time from the point at which the MedSpect achieved 10% of its ¢nal response to the analgous points in the RAMS's response.
Part II Animal preparation The use of dogs in this protocol was approved by the Institutional Animal Care and Use Committee. Four female beagle dogs (weight 9.0 kg to 9.6 kg) were fasted 12 hr immediately prior to the experiment. They were premedicated with 0.5 mg kgÿ1 of acepromazine subcutaneously. Anesthesia was induced with pentobarbital sodium (30 mg kgÿ1 ) via IV bolus. Maintenance IV anesthetic consisted of thiopental sodium (3 mg kgÿ1 hrÿ1 ) and pancuronium bromide (0.08 mg kgÿ1 hrÿ1 ). During periods when inhaled anesthetics were used, the infusion rates of thiopental and pancuronium were halved, and 1 mg IV boluses of pancuronium were administered hourly. The dogs were placed in a supine position, intubated with an 8.0 mm ID cu¡ed endotracheal tube (Hi-Lo Jet, Mallinckrodt, Glen Falls, NY), and connected to a servo-ventilator (Siemens 900C, Sweden) with minute ventilation adjusted to maintain end-tidal CO2 between 35 and 40 mmHg at 12 breaths minÿ1 . Arterial oxygen saturation was monitored continuously by pulse oximeter (Criticare 504-US, Waukesha WI) while body temperature was maintained between 35 and 37 ³C with a heat lamp and blanket. Anesthetic gases were supplied via an anesthesia machine with separate vaporizers (Cyprane Ltd, England) for iso£urane, halothane and en£urane. Data acquisition The mass spectrometers were calibrated as described above and checked against the calibration mixture on an hourly basis. The same optimal catheter con¢gura-
183
tion as in Part I was used for the MedSpect. The analog outputs for O2 , N2 O, CO2 , halothane, iso£urane, and en£urane were recorded by a 386 computer via an analog-to-digital converter (Computer Boards Inc. #CIO-AD16, Mans¢eld MA) sampling at 18.2 Hz. Manufacturer-supplied software calculated, in realtime, inspired and end-expiratory values for each gas as well as respiratory rate. The RAMS was ¢tted with the standard sampling line (10-ft, 0.062-in ID) without any water trap in place, and sampled at 240 ml minÿ1 . It sent digital data at a rate of 20 Hz to the computer through its serial data port. Values for inspired and endtidal were calculated by the same manufacturer-supplied program. Inspired and end-tidal values for all gases from both mass spectrometers were recorded into a permanent data ¢le for subsequent data analysis. In addition, CO2 waveforms from both mass spectrometers were continuously recorded on the chart recorder at a speed of 5 mm secÿ1 . The sampling catheter tips were at the same location, in the center of the £ow between the endotracheal tube and the ventilator tubing. Protocol Five di¡erent mixtures of gases (Table 2) were used in random order for ventilating, and the dogs were ventilated with 100% O2 for 20 min before using each anesthestic gas mixture. For each gas mixture, the respiratory rate on the ventilator was systematically changed to 8, 20, 12, 15 and 10 breaths minÿ1 while holding minute ventilation constant. Measurements using inspiratory/expiratory ratios (I : E) of 1 : 1 and 1 : 2 were made at each respiratory rate. This gave 50 combinations of anesthetic and ventilator settings per dog, a total of 2,400 di¡erent points of comparison between the mass spectrometers. Following measurements in this fashion, the dogs Table 2. Gas mixtures (volume %) used for modeling typical anesthesia in beagle dogs. Totals do not add up to 100% due to uncertainties in the O2 and N2 O values of up to 2% Gas O2 mixture
N2 O
Halothane
En£urane
Iso£urane
1 2 3 4 5
50% 50% 70% 30% 30%
1.0% 2.0% ^ ^ ^
^ ^ 1.5% ^ ^
^ ^ ^ 0.5% 1.0%
50% 50% 30% 70% 70%
184
Journal of Clinical Monitoring Vol 13 No 3 May 1997
Table 3. 10^90% rise times (in sec) of the MedSpect and RAMS for O2 , CO2 , N2 O, and iso£urane for the three catheter con¢gurations listed in Table 1. Also listed is the delay (sec) of the RAMS relative to the MedSpect. Iso = iso£urane. All values for the RAMS may be underestimates by as much as 0.045 sec due to the stepwise nature of its output RAMS catheter
Stainless steel catheter Standard sampling catheter
Water trap? No No Yes
MedSpect rise times O2
CO2
N2 O
Iso
O2
CO2
N2 O
Iso
Delay of RAMS CO2
0.11
0.11
0.11
0.12
0.07 0.16 0.24
0.07 0.16 0.30
0.07 0.16 0.24
0.05 0.24 0.30
0.19 1.10 1.17
were allowed to awaken while being ventilated with 100% O2 . Data analysis The inspiratory and end-tidal values of each measured gas were calculated by averaging the last 5 breaths of the 5-min data collection period. The methods of Lee et al. [13] were used to assess the interchangeability of the two mass spectrometers' results. With this approach, two measurement techniques are considered interchangeable if the lower bound of the 95% con¢dence interval, q, of the intraclass correlation is greater than 0.75 and if the mean di¡erence between the two is insigni¢cant (p > 0.05, as determined by ANOVA with Bonferroni correction). These calculations were repeated for each measured gas at each respiratory rate and I : E. Lee et al. [13] point out that to use q is to assume the lack of interaction terms and further state that when
RAMS rise times
applied to data lacking repeated measures, such interaction terms can only be detected by a graphic method. The present study did employ repeated measures, so the signi¢cances of the interaction terms were determined statistically (using p < 0.05 to denote signi¢cance) using the general linear model described by Lee et al. [13], and used to verify the applicability of using their methods. For those gases demonstrating clinically signi¢cant di¡erences in the values reported by the two mass spectrometers via the ANOVA, we tested whether these di¡erences were correlated with respiratory rate using linear regression. We also used linear regression to further test whether the increased di¡erences were correlated to higher measured gas concentrations. The accepted level of signi¢cance in these regressions was 0.05. RESULTS
Calibration Over the course of 7^8 hours of continuous use, the MedSpect typically required recalibration after 3^4 hours. The RAMS did not require recalibration over the period of a week.
Part I
Fig. 1. Responses of the MedSpect and RAMS to a 0^62% step change in CO2 : a) MedSpect using stainless steel catheter at 60 ml minÿ1 ; b) RAMS using stainless steel catheter at 60 ml minÿ1 ; c) RAMS using standard sampling catheter at 240 ml minÿ1 ; and d) as in ``c'' with addition of AQUA-Knot 2. The RAMS updates its analog outputs every 45 milliseconds resulting in the illustrated staircase e¡ect.
The variation in measured rise and lag times under static conditions were within measurement error ( 0.01 sec for the MedSpect and 0.045 sec for the RAMS). Figure 1 illustrates the typical responses of both mass spectrometers to a 0^62% step change in N2 O. Table 3 summarizes the 10^90% response times and relative delays of the MedSpect and RAMS for each catheter con¢guration listed in Table 1. The MedSpect's rise times for O2 , CO2 , N2 O and iso£urane are 0.11^0.12 sec when using the stainless steel catheter at 60 ml minÿ1 . When ¢tted with an identical catheter and set to
Delaney et al: ResponseTime Studies of a New, Portable Mass Spectrometer
185
Fig. 2. Typical capnograms of each mass spectrometer at a low (8 breaths minÿ1) and high (20 breaths minÿ1) respiratory rates.
sample at the same £ow rate the RAMS had rise times of 0.05^0.12 sec, with a delay of 0.19 sec. Using a standard sampling catheter at 240 ml minÿ1 without the AQUA-Knot 2 , the RAMS's rise times were 0.16^0.24 sec with a delay of 1.10 sec. With the addition of an AQUA-Knot 2 the rise times in the RAMS were 0.24^0.30 sec with a delay of 1.17 sec.
Part II Figure 2 illustrates capnograms for the MedSpect and the RAMS at 8 and 20 breaths minÿ1 . Figure 3 shows the di¡erences between the mass spectrometer readings for each inspired gas measurement versus the mean of both mass spectrometers at each point, while Figure 4 shows the analogous data for end-tidal measurements. Table 4 summarizes the statistically signi¢cant results of the analyses. In many of the plots, data fall into distinct clusters corresponding to the di¡erent mixtures of inspired gases (Table 2). For all variables, the interaction term described by Lee et al. [13] was found to be insigni¢cant (p > 0.05), indicating the validity of applying their analysis methods to the data in this study. There were no di¡erences between the values reported by the two devices (p > 0.05 and q > 0.75) for either inspired or end-tidal N2 O, halothane, iso£urane, and O2 regardless of respiratory rate and I : E ratio (with the exception of inspired O2 at 20 breaths minÿ1 and I : E = 1 : 2). The RAMS gave
a higher end-tidal en£urane value (p < 0.05) for all but one ventilator setting (20 breaths minÿ1 at 1 : 2), with a maximum di¡erence of 0.035%. Di¡erences for inspired en£urane were also found (p < 0.05), up to 0.025%. For respiratory rates up to 15 breaths minÿ1 with I : E = 1 : 2, the RAMS reported a higher value for inspired CO2 (p < 0.05) by up to 0.089% (0.63 mmHg). The di¡erence between inspired CO2 as measured from the MedSpect and RAMS increased with mean inspired CO2 (q < 0.75). In measurements of end-tidal CO2 , the RAMS reported values lower than the MedSpect (p < 0.05) with mean di¡erences increasing with mean CO2 (q < 0.75). The greatest mean di¡erence between the MedSpect and RAMS was 0.49% (3.7 mmHg). The highest values for the di¡erences in end-tidal CO2 between the MedSpect and RAMS occurred at the higher respiratory rates. Only the mean di¡erences in end-tidal CO2 were clinically signi¢cant (Table 4), and linear regression revealed that these di¡erences increased with increasing respiratory rate (p < 0.05) and with increases in the average of the end-tidal CO2 reported by the MedSpect and RAMS (p < 0.05). At none of the 200 measured points did the respiratory rates calculated by the MedSpect and the RAMS di¡er by more than 0.1 breaths minÿ1 , the resolution of each mass spectrometer.
186
Journal of Clinical Monitoring Vol 13 No 3 May 1997
Fig. 3. Plots of di¡erences between the RAMS and MedSpect values versus their mean values for inspired O2 , N2 O, CO2 , halothane, en£urane and iso£urane. Shapes of symbols indicate the respiratory rate. These are, respectively, for 8, 10, 12, 15 and 20 breaths minÿ1 : *, !, ~, &, and ^. Black symbols are for I : E = 1 : 2, white symbols for 1 : 1. For each plot, n = 200.
DISCUSSION All mass spectrometers have a delay in addition to that associated with trans-catheter transport of the samples. This intrinsic delay is a property of the mass spectrometer and has many causes, including transport time from the mass spectrometer's inlet to the ionization chamber and subsequent electronic processing. The results of Part I, when both devices had the same catheter con¢gurations, indicate that the RAMS has an intrinsic delay of 0.19 sec more than that of MedSpect. This delay did not adversely a¡ect the RAMS's rise times, which were lower (Table 3) than those of the MedSpect when
operating with an identical catheter and £ow rate. However, it must be noted that due to the stepwise nature of the RAMS's analog output which is updated every 0.045 sec (Figure 1), the accuracy of these rise times is unclear, and they may in fact be as high as the values seen in the MedSpect. Nonetheless, the RAMS did at least as well as the MedSpect in this regard. Switching the RAMS to the larger and longer catheter resulted in longer rise times (by 0.10 sec) and in the delay (by 0.91 sec), phenonema well-documented in the literature [4^8]. The increase in catheter transit time allows more time for di¡usion of the samples, lengthening the rise times. Sampling at 240 ml secÿ1 through
Delaney et al: ResponseTime Studies of a New, Portable Mass Spectrometer
187
Fig. 4. Plots of di¡erences between the RAMS and MedSpect versus their mean values for end-tidal O2 , N2 O, CO2 , halothane, en£urane and iso£urane. Symbols same as in Figure 3.
a standard sampling catheter gives a calculated mean transit time of 1.63 sec (compared to 0.007 sec for the smaller catheter). However, calculating the Reynolds number for each catheter reveals that £ow in both is laminar, implying that the maximum £ow, found along the central axis of the catheter, is twice the mean £ow [7]. Therefore the very ¢rst traces of a given sample require 0.81 sec to reach the RAMS's inlet port. Adding in the 0.19 sec intrinsic delay of the RAMS gives a total estimated delay of 1.00 sec, very close to the measured value of 1.10 sec. Were the intrinsic delay of the RAMS mostly due to internal dead space it could be expected to decrease with higher £ow rates. That no di¡erence in the RAMS's
intrinsic delay of 0.19 sec was seen between 60 and 240 ml minÿ1 indicates that it cannot be easily remedied. However, since this intrinsic delay is not associated with di¡erent rise times, inspired and end-tidal values should not be a¡ected. It should be noted that the RAMS used in this study was the version designed for clinical use. A di¡erent RAMS designed for research purposes, also available from the manufacturer, may have di¡erent delays and rise times. When used clinically, samples to be analyzed are ¢rst sent through a water trap to prevent damage to mass spectrometers. Since these devices increase the dead space through which the samples travel they exacerbate the distortions associated with sampling catheters.
188
Journal of Clinical Monitoring Vol 13 No 3 May 1997
Table 4. Statistically signi¢cant (p < 0.05) mean di¡erences (in percent) between values for the measured gas concentrations from the RAMS and MedSpect. Positive di¡erences indicate that higher values were given by the RAMS. When the lower bound of the intraclass correlation's 95% con¢dence interval (q) was less than 0.75 its value is given in parentheses. Only if mean di¡erences are insigni¢cant and q > 0.75 are results from the two di¡erent sources considered interchangable.Values not listed are insigni¢cant (p < 0.05 and q > 0.75) Gases
Breaths per minute/inspiratory : expiratory ratios 8/1:2
Inspiratory CO 2 En£urane
0.032 (0.017)
10 / 1 : 2
12 / 1 : 2
15 / 1 : 2
20 / 1 : 2
8/1:1
10 / 1 : 1
12 / 1 : 1
15 / 1 : 1
20 / 1 : 1
0.071 (<0.001)
0.087 (<0.001)
0.089 (<0.001)
(<0.001)
(<0.001)
(<0.001)
(<0.001)
0.049 (<0.001)
(<0.001)
0.022
0.025
End-tidal O2 CO2 En£urane
0.025
0.709 ÿ0.326 (0.095)
ÿ0.292 (0.421)
ÿ0.274 (0.543)
ÿ0.328 (0.401)
0.028
0.032
0.033
0.035
The AQUA-Knot 2 used in this study has a dead space of 0.6 ml [14], so at 240 ml minÿ1 can be expected to lengthen any delays by an additional 0.15 sec, or half this if £ow through the AQUA-Knot 2 is laminar. Our measured value of 0.07 sec implies this to be the case. At the same time, the AQUA-Knot 2 increased the 10^90% rise times by 0.08^0.14 sec. Such distortions could be expected to degrade the RAMS's inspiratory and expiratory measurements, particularly at high frequencies. In Part II, we found several small but statistically signi¢cant di¡erences between the values reported by the MedSpect and RAMS. For en£urane the greatest mean di¡erence was only 0.035% (0.25 mmHg), well within limitations of instrument calibrations. Di¡erences seen for inspired CO2 and for end-tidal CO2 at the low respiratory rates are also within these limits. Linear regression revealed that end-tidal CO2 di¡erences were correlated with respiratory rate and mean end-tidal CO2 levels. This latter e¡ect means that these di¡erences could have increased just due to the fact that the absolute values for end-tidal CO2 were higher. Thus, a small calibration error would be exaggerated at high CO2 , increasing the end-tidal CO2 di¡erence. Alternatively, the increased di¡erences could have been caused by the di¡erent response times of the two devices at higher respiratory rates. However, since end-tidal CO2 increased at the higher respiratory rates (a consequence of the increased dead space associated with using higher frequencies at a constant minute ventilation), it cannot be determined whether the increased CO2 di¡erences were due to discrepancies in instrument
ÿ0.486 (0.192)
ÿ0.339 (0.164)
ÿ0.287 (0.516)
ÿ0.262 (0.581)
ÿ0.364 (0.460)
ÿ0.421 (0.415)
0.031
0.035
0.031
0.036
0.035
calibration, to use of the larger catheter in the RAMS, or to a combination of both. In any case, these factors only distort end-tidal CO2 values to a small degree (e.g., the di¡erence being 3.7 mmHg at 20 breaths minÿ1 ) which under most clinical circumstances are inconsequential. Therefore, the RAMS when used with a catheter and calibrated as it would be clinically, provides generally the same results as a standard mass spectrometer used with an optimum catheter and calibrated frequently. In clinical practice, mass spectrometers like the MedSpect are rarely used in the idealized conditions employed in this study. Rather they are typically used in a timesharing fashion where they analyze samples from as many as 28 di¡erent sites [11] via lengthy catheters which degrade their accuracy [4^8]. Therefore under ordinary clinical conditions a MedSpect would not give the accuracy seen in this study. From this we conclude the RAMS o¡ers a level of accuracy comparable, if not superior, to that of the MedSpect when used clinically, and eliminates factors that degrade the performance of time-shared systems. The authors wish to acknowledge the contributions of Rene¨ e Kahn for her assistance with the animals, and Charles Wilkinson who provided technical assistance with both mass spectrometers. This study was funded in part by a grant from Marquette Electronics, Inc.
Delaney et al: ResponseTime Studies of a New, Portable Mass Spectrometer
REFERENCES 1. Frazier WT, Odom SH. E¤ciency and expense of timeshared mass spectrometer systems. Biomed Instrum and Technol 1989; 23: 481^484 2. Paulsen AW. Spare mass spectrometer vs. linking systems in the event of a single system failure. J Clin Monit 1992; 8: 319^320 3. Steinbrook RA, Elliott WR, Goldman DB, Philip JH. Linking mass spectrometers to provide continuous monitoring during system failure. J Clin Monit 1991; 7: 271^273 4. Carlon GC, Kopec IC, Miodowik S, Ray C. Frequency response of the peripheral sampling sites of a clinical mass spectrometer. Anesthesiology 1990; 72: 187^190 5. Paulsen AW. Factors in£uencing the relative accuracy of long-line time-shared mass spectrometry. Biomed Instrum Technol 1989; 23: 476^480 6. Turner JC. Use of long catheters for multipatient anesthetic monitoring at high respiratory frequencies. J Clin Monit 1991; 7: 237^240 7. Tavener SJ,Withy SJ, Harris EA. Response characteristics of a mass spectrometer. Med Bio Engin Comput 1984; 22: 493^498 8. Lerou JG, van Egmond J. Mass spectrometry: Performance of long catheters. J Clin Monit 1993; 9: 68^69 9. Schulte GT, Block FE. Evaluation of a single-room, dedicated mass spectrometer. Int J Clin Monit Comput 1991; 8: 179^181 10. Beatty PC. The Spectralab-M quadrupole medical mass spectrometer. J Med Engin Technol 1988; 12: 265^272 11. SARA Service Manual. PPG Biomedical Systems 1987 12. Sodal IE. The medical mass spectrometer. Biomed Instrum Technol 1989; 23: 469^476 13. Lee J, Koh D, Ong NC. Statistical evaluation of agreement between two methods for measuring a quantitative variable. Comp Bio Med 1989; 19: 61^69 14. AQUA-Knot 2 Technical Speci¢cations. Marquette Electronics Incorporated, 1991.
189
