Accuracy of Postoperative End-tidal Pco²Measurements with Mainstream and Sidestream Capnography in Non-obese Patients and in Obese Patients with and without Obstructive Sleep Apnea

The University of Louisville Human Studies Committee (Louisville, Kentucky) approved the protocol, and each patient gave written informed consent. We recruited 60 patients who were scheduled for general anesthesia with continuous arterial pressure monitoring via  an arterial catheter; 20 were non-obese (defined by a body mass index less than 30 kg/m²) without a diagnosis of OSA (Non-obese non-OSA), 20 were obese (body mass index greater than 35 kg/m²) without a diagnosis of OSA (Obese non-OSA), and 20 were obese with OSA diagnosed by polysomnography (Obese OSA).

To avoid assigning undiagnosed OSA patients in either of the non-OSA groups, patients were asked to complete the Epworth Sleepiness Scale,20a simple questionnaire to determine daytime sleepiness. Patients having an Epworth Sleepiness Scale of 10 or more were excluded from the non-OSA groups. Patients with known severe pulmonary disease (Hugh-Jones classification grade 3 or above), cardiac disease (New York Heart Association classification grade 3 or above), or who would be indicated for a facemask for postoperative oxygen delivery rather than nasal cannula were also excluded.

General anesthesia was administered by using tracheal intubation or a laryngeal mask airway, with no other restrictions on anesthetic management.

Patients were extubated in the operation room and just after patients admitted to PACU, patients were randomly assigned to three different capnometer systems (device/cannula combination) (fig. 1): (A) Cap-ONE mainstream capnometer system (Nihon Kohden) that includes an oral guided nasal cannula (Mainstream oral guide); (B) Microcap sidestream capnometer with a Smart CapnoLine Plus nasal cannula (Oridion Capnography Inc.) that incorporates an oral guide along with a carbon dioxide sampling port (Sidestream oral guide); or (C) Microcap sidestream capnometer (Oridion Capnography Inc.) with a CapnoLine H nasal cannula (standard nasal cannula with no oral guide) (Sidestream standard).

All patients had their Etco²measured by using each system and a constant oxygen flow rate of 4 l/min through the nasal cannula. Because the respiratory state was not always stable during the measurements and might be influenced by opioids given for pain control, the application order of the two capnometers and two sidestream cannulas was randomized on the basis of computer-generated assignments that were kept in sequentially numbered opaque envelopes that were opened in the PACU.

Exhaled carbon dioxide was recorded (capnogram) for 5 min by using each capnometer system. At the end of each 5-min interval, an arterial blood sample was taken for blood gas analysis. (GEM Premier 3000; Instrumentation Laboratory, Lexington, MA).

Morphometric and demographic characteristics of the participating patients were recorded, i.e. , age, body weight, height, body mass index, gender, race, length of philtrum, and type of surgery. Etco²was determined breath-by-breath by each capnograph. The average Etco²during the final 60 s of 5-min measurement period was recorded by a computerized system.

During measurements, if needed hydromorphone or morphine was given for analgesia according to physician's order who was independent of this study.

Our major outcome was the accuracy of Etco²measurements with each capnometer system in the three patient groups. The mean Etco²value for each device/cannula combination measurements was subtracted from the arterial partial pressure of carbon dioxide (Paco²) measurement that was measured simultaneously. This provided the difference between the Etco²and the Paco²(Δco²), which defined accuracy.

An average Δco²difference of 3 mmHg between any two combinations of capnometers and nasal cannulas in obese versus  non-obese or OSA versus  non-OSA was determined a priori  as technically important bias. From a preliminary study of the mainstream device (cap-ONE; Nihon Kohden), we expected the SD of the Δco²to be 4.5 mmHg. Assuming values within patients to be correlated at least at r = 0.5, we needed 17 patients in each patient group to have 90% power to detect a difference of 3 mmHg with a repeated-measures ANOVA, α= 0.05.

Our secondary outcome was oxygenation efficacy. Etco², Δco², and Pao²were compared in a two-factor (both having three levels), mixed model, repeated-measures ANOVA. The factors were (1) the three device/cannula combinations (Mainstream oral guide, Sidestream oral guide, and Sidestream standard) and (2) the three patient groups (Non-obese non-OSA, Obese non-OSA, Obese OSA). Significant differences between groups were analyzed using Tukey post hoc  testing. To further analyze the effectiveness of each combination of capnometer and nasal cannula, we examined Pearson's correlation coefficient and used a Bland-Altman analysis to check for bias and effect modification with the capnometer and nasal cannula combinations across the full range of measured values.21 

Data are presented as means ± SDs for continuous variables and percentages for categorical variables. P < 0.05 was considered statistically significant. All statistical analyses were done by SPSS for Windows version 16.0 (SPSS Inc, Chicago, IL).

A total of 180 Etco²–Paco²measurement pairs were analyzed from 60 patients. Of these, 20 patients were non-obese without OSA (Non-obese non-OSA), 20 were obese without OSA (Obese non-OSA), and 20 were obese with OSA (Obese OSA). Patient characteristics for the three groups are shown in table 1. Although gender was not uniformly distributed among the groups, the length of the philtrum was similar in all groups. Patients in the Obese OSA group reported a significantly greater Epworth Sleepiness Scale score (10 ± 3) than the Non-obese non-OSA group (2 ± 2) and the Obese non-OSA group (2 ± 2; P < 0.001). No patients in the Non-obese non-OSA or Obese non-OSA groups reported a score greater than 10.

There were significant differences in Δco²among the three device/cannula combinations in all three patient groups. Δco²was smallest when measured with the mainstream capnometer, slightly greater with the sidestream capnometer with an oral guide, and greater still for the sidestream capnometer with the standard nasal cannula (fig. 2).

Pao²was similar in all groups during all device/cannula combinations, except for the obese non-OSA patients who had a higher Pao²during the period when they were measured by sidestream standard device (fig. 3). All other blood gas measurements, including pH, bicarbonate, base excess, and Paco², were similar among device/cannula combinations in all patient groups.

The correlations between Etco²and Paco²are shown in figure 4. Correlation coefficients were highest with the Mainstream oral guide device and lowest with the Sidestream standard device, but they tended to be highest in Non-obese non-OSA group and lowest in Obese OSA group. Carbon dioxide data and correlations were the most accurate when using the mainstream device.

The bias in the relationship between Etco²and Paco²is illustrated in figure 5where the difference is plotted against the average of the values.21On Bland-Altman plots, bias was more widely distributed in Obese patients with and without OSA. Bias was also greater with the standard nasal cannula than with the cannula that included an oral guide.

This study shows that Etco²measured with a mainstream capnometer produces better correlation between Etco²and Paco²than other systems. Our results also indicate that both obesity and OSA reduce the correlation between Etco²and Paco²and accordingly increase Δco²unpredictably. Bland-Altman plots show that the mainstream capnogram was more accurate in predicting Paco²than the sidestream system, especially in patients with obesity and OSA.

In healthy young adults, Δco²is normally between 2 and 5 mmHg.22,23Δco²tends to increase with age, obstructive lung disease, in situations where alveolar dead space increases24and results in a ventilation-perfusion mismatch, and with hemodynamic instability as well. Ventilation and perfusion mismatch is commonly increased immediately after endotracheal tube extubation because of transient atelectasis. Obesity is also considered to be a risk factor for postoperative atelectasis.25Obese patients are therefore prime candidates for having a large Δco²in during recovery. Takano et al.  26showed that Δco²is highly dependent on tidal volume. Δco²is expected to be high when the tidal volume is near dead space volume or when the respiratory rate gets too high to give an end-expiratory plateau, or patient's airway is partially obstructed due to bronchospasm.27 

The determined Δco²values in this study were greater than those previously reported. Bowe et al.  9reported a Δco²of 2.1 ± 2.2 mmHg and a Paco²of 38.6 ± 3.8 mmHg by using a sidestream nasal cannula capnometer in preanesthetic patients getting 3 l/min oxygen. The most likely reason for the differences between the reported values is the dissimilarity of the circumstances under which the data were collected. Bowe et al.  obtained their data before induction of anesthesia, whereas our results were obtained during recovery. This is an important distinction because hypoventilation or mouth breathing diminishes nasal expiratory flow rate, thereby increasing Δco².

Etco²is defined as partial pressure of carbon dioxide at the end of the expiratory phase of the respiratory cycle. However, dilution and a physiologic obstructed pattern often results in inconsistency of the expiratory plateau and slope of Phase 3 of the Etco²waveform.28Consequently, Etco²values are especially misleading under the combination of severe obstructive pattern, low expiratory flow, low peak flow rates, and high fresh oxygen supply.

Given that the oxygen supply rate of 4 l/min was constant for all patients, the inconsistencies of Δco²between patient groups likely resulted from differences in dead space and respiratory pattern for each patient group. Our results demonstrate that the arterial-to-end-tidal Pco²gradient in spontaneously breathing postoperative patients depends on the patient population and type of capnometer. Oral guide devices proved to be more accurate measure of Etco²as compared to a standard sidestream measurement. In non-obese and obese patients without OSA, readings from mainstream capnometers were both accurate and statistically significantly better than alternative approaches.

We note, though, that even in obese OSA patients—the most vulnerable patient population29—the difference of mean Δco²between mainstream device and sidestream with conventional cannula was only 4.3 mmHg. This difference is relatively small because many clinicians already assume that Etco²underestimates arterial partial pressures by 2–5 mmHg. But an important factor is that the correlation between Etco²and Paco²was low without an oral guide in obese OSA patients (r = 0.39) and was markedly improved by addition of an oral guide (r = 0.72). Considering that OSA patients are at the high risk for adverse respiratory events and that intense respiratory monitoring is recommended, our results support using a cannula that includes an oral guide for OSA patients.

The cost of these systems is difficult to estimate because there are differences from country-to-country, and cost depends on use levels and negotiating power of specific hospitals. However, it appears that the nondisposable and disposable components for each tested system cost similar amounts. To the extent that this proves to be the case in any particular hospital, clinicians will presumably prefer the most accurate system.

In summary, mainstream capnometry performed best, and an oral guide improved the performance of sidestream capnometry. Accuracy in non-obese and obese patients, with and without OSA, was similar.

The authors thank Gilbert Haugh, M.S., Biostatistician, Office of Clinical Research Services and Support, University of Louisville, Louisville, Kentucky, for help with statistical analysis.