Secondary Logo

Journal Logo

General Articles: Research Report

Capnogram Shape in Obstructive Lung Disease

Krauss, Baruch MD, EdM*‡; Deykin, Aaron MD†‡; Lam, Alexander; Ryoo, Joan J.; Hampton, David R. PhD§; Schmitt, Paul W. PhD§; Falk, Jay L. MD

Author Information
doi: 10.1213/01.ANE.0000146520.90393.91
  • Free

Capnography, the noninvasive measurement of the partial pressure of carbon dioxide (CO2) in exhaled breath, provides a real-time assessment of ventilatory status (1). It also provides a graphic representation of CO2 concentration over time represented by the CO2 waveform (capnogram), whose shape relates to the airflow and emptying characteristics of the lung during each breath (2). In subjects without respiratory difficulty or underlying lung pathology, the capnogram has a rectangular shape, with the CO2 concentration rapidly increasing on exhalation to a steady value represented by the alveolar plateau. The capnogram achieves a maximal value at the end of exhalation (the end-tidal CO2 [ETco2] concentration) then decreases rapidly to zero as inhalation begins (Fig. 1) (1–4).

Figure 1.:
The normal time-based capnogram with phases A–E and take-off and elevation angles marked.

In small, preliminary studies, specific changes in capnogram shape were observed in obstructive lung disease (OD). The capnogram had a more rounded appearance during the initial phase of exhalation and an upward slope during the alveolar plateau. These changes were correlated to forced expiratory volume in 1 s (FEV1) (5–11). The primary objective of this study was to validate these observed changes in capnogram shape on a larger dataset using well-defined continuous-valued measurements of relevant features. This enabled us to rigorously distinguish capnograms of subjects with spirometry-defined OD from those of subjects with similarly defined restrictive lung disease (RD) and those of normal subjects. The secondary objective was to validate the observation that changes in capnogram shape in OD correlate with changes in FEV1.


The study was approved by the Brigham and Women’s Hospital (Boston, MA) IRB. Written informed consent was obtained from subjects before entry into the study. Confidentiality was protected by identifying each record with an indirect study identifier: no information could be used to directly identify the study participants.

The study was conducted in a prospective, nonrandomized manner using a convenience sample of spontaneously breathing subjects, presenting to a pulmonary function laboratory for scheduled pulmonary function testing (PFT) from June 2001 to September 2002. Normal subjects, nonsmokers without lung disease, were recruited as controls. PFT was done in accordance with American Thoracic Society standards (12). Before starting PFT, subjects were connected, by an oral-nasal cannula (sampling CO2 from the nose and the mouth), to a portable multiparameter monitor (Medtronic LIFEPAK®12, Redmond, WA) with low-flow sidestream (Microstream®) capnography (13). The capnograph collects a continuous sample of air at a flow rate of 50 mL/min and records the instantaneous CO2 concentration every 40 ms. A 90-s sample, containing at least 8 spontaneous breaths, was collected from each patient. At the conclusion of the recording, all data were transmitted to a desktop computer for analysis.

Standard spirometric measurements from PFT, including FEV1 and forced vital capacity, obtained immediately after the capnography recording, were interpreted by an independent pulmonologist to confirm the patient’s diagnostic category and their degree of impairment in pulmonary function. Patient diagnosis was confirmed as being OD, RD, or normal using the FEV1/forced vital capacity ratio (RD: ratio >0.75; OD: ratio ≤0.75). Severity of the patient’s breathing impairment was determined based on standard spirometric criteria: normal (FEV1 >80% of predicted value), moderate (FEV1 ≤80% and >50% of predicted value), or severe (FEV1 ≤50% of predicted value) (Table 1). Thus, subjects were categorized into one of four groups: normal, moderate obstructive, severe obstructive, or restrictive.

Table 1:
Diagnostic Criteria

Spirometric and capnography data were linked together in a database so that an independent, objective assessment could be made of the relationship between the derived capnography measures and the patient’s diagnosis and severity. A computer algorithm, able to automatically identify and consistently measure the major features of the capnogram, was used to process the raw data according to the following steps. First, criteria of slope and amplitude were used to detect turning points in the CO2 concentration, corresponding to the start of inhalation and the end of exhalation, that defined each breath. Within a breath, the linear phases of the capnogram were identified, including the initial expiratory rise and alveolar plateau. Corresponding slopes for each feature, specified in (mm Hg/s) units, were computed from straight lines regressed through the sample points. Finally, associated angles from the horizontal were calculated, including the take-off angle of the initial expiratory rise, and the elevation angle for the slope of the alveolar plateau. Supplemental measures, consisting of the ETco2 concentration, respiration rate, and inspiratory and expiratory times, were also determined to give a total set of six characteristic values for each breath (ETco2 value, respiration rate, take-off angle, elevation angle, inspiratory time, expiratory time). Median values for each of these six capnogram features were determined across the set of recorded breaths for each patient, lessening the influence of outliers. The median values, in turn, were averaged over all subjects belonging to the same ventilatory group and the statistical significance of all paired ventilatory group differences was assessed using the Tukey-Kramer multicomparison procedure.


Two-hundred-sixty-two subjects were enrolled. Of these subjects, one patient was excluded from analysis of clinical status because the PFT data were lost. Two-hundred-seven (79%) were Caucasian, 108 (41%) were between 55 and 70 yr of age, and 150 (58%) were female (Table 2). One-hundred-ten (42%) subjects were categorized by diagnosis as OD, 91 (35%) as normal, and 60 (23%) as RD (Table 3).

Table 2:
Table 3:
Descriptions and Definitions by Patient Category

Capnograms derived from subjects with OD differed from those obtained from normal and RD subjects (Fig. 2). Restrictive subjects had capnograms that were similar to normal subjects in all six of the characteristic measures. Among the six quantitative measures (ETco2 value, respiration rate, take-off angle, elevation angle, inspiratory time, expiratory time), the most marked differences were in the angles of the initial expiratory rise and of the alveolar plateau (the take-off angle and the elevation angle, respectively; Fig. 1). These differences were progressive, increasing with severity as documented by a decreasing FEV1 (Figs. 3 and 4). For both the take-off and elevation angles, we were able to demonstrate, using the Tukey-Kramer multicomparison procedure, that statistically significant differences between the mean of the severe OD group and all other group means (moderate obstructive, normal, and RD) existed at the 5% significance level (Figs. 3b and 4b).

Figure 2.:
Capnogram shape in normal, obstructive, and restrictive subjects. Each capnogram is aligned with its respective forced expiratory volume in 1 s (FEV1) and ratio of FEV1/forced vital capacity (FVC) on the left. The top capnogram is from a normal patient, the middle capnogram from a patient with severe obstructive lung disease, and the bottom capnogram from a patient with restrictive lung disease.
Figure 3.:
Take-off angle for each ventilatory condition with confidence intervals. A, take-off angle versus ventilatory status. B, paired ventilatory differences, take-off angle. sO = severe obstructive lung disease (OD), mO = moderate OD, n = normal, r = restrictive lung disease.
Figure 4.:
Elevation angle for each ventilatory condition with confidence intervals. A, elevation angle versus ventilatory status. B, paired ventilatory differences, elevation angle. sO = severe obstructive lung disease (OD), mO = moderate OD, n = normal, r = restrictive lung disease.

The average take-off angle of the ascending phase of the capnogram for the severe OD group was 7.2 degrees less (95% confidence interval [CI]: 4.0, 10.4) than the take-off angle for the normal group. The average elevation angle of the alveolar plateau was 0.8 degrees more (95% CI: 0.14, 1.4) for the moderate OD group than for subjects with normal spirometry; whereas the average elevation angle was 3.6 degrees more (95% CI: 2.9, 4.3) for the severe OD group than for subjects with normal spirometry (Figs. 3b and 4b).


The shape of the capnogram in normal subjects and subjects with OD has been the subject of investigation for >40 years. Early studies, in the 1960s, focused on characterizing the shape of the capnogram in normal subjects (3), on demonstrating differences between their capnograms and capnograms from OD subjects (4,8,9), and on determining the physiological mechanisms underlying the capnogram changes in OD (3,5–7). More recent studies, in the mid-1990s, demonstrated a correlation between FEV1 and the shape of the capnogram in adult asthma subjects (10,11). However, all of these were small, preliminary studies, most with <30 OD subjects. A larger dataset was needed to validate the findings in these studies.

We sought to validate these intriguing observations by using a more rigorous methodology that included well-defined, continuous-valued measurements of the relevant capnogram features. We developed precise criteria for identifying and measuring the features that we thought were related to changes observed in OD. These quantitative measures of the capnogram could take on any numeric value, as opposed to categorical or binary (present/absent) categorizations, and thus could be evaluated using descriptive statistics. Whereas prior studies had noted changes that seemed to be associated with OD, we took these observations as a starting hypothesis and created a mathematical description of the features, prospectively collected a large dataset, and evaluated the measures to establish statistical distinctions between the groups.

We studied subjects with OD in order to characterize the capnograms of this group of subjects in comparison to capnograms from normal and RD subjects. Changes in capnogram shape previously observed in OD were confirmed and the correlation between changes in capnogram shape and changes in FEV1 was validated. Significant differences were noted between capnograms of subjects with spirometry-defined OD as compared with those derived from subjects with spirometry-defined RD and those of subjects with normal spirometry.

The initial, rapid increase in CO2 concentration reflects the time required for the dead space volume to pass across the CO2 sensor. At normal flow rates, this 150-mL volume passes in 250 ms, resulting in a sharp upstroke on the capnogram (B–C in Fig. 1). Diminished flow rates in OD prolong the transition as the dead space volume takes longer to be exhaled, causing the slope and take-off angle to diminish in direct proportion to the severity of the obstruction.

The plateau phase of the capnogram (C–D in Fig. 1) reflects the passage of air from progressive emptying of the alveoli. Normally, alveoli are equally ventilated and all have similar CO2 concentrations, so the capnogram has a nearly constant value throughout this phase of exhalation. In OD, some terminal bronchi are narrowed, resulting in local hypoventilation of alveoli and an associated increase in their CO2 concentration. Alveoli attached to unobstructed terminal bronchi are relatively hyperventilated and have diminished CO2 concentrations. During exhalation, unobstructed alveoli empty ahead of obstructed ones, leading to a progressive increase in CO2 concentration during exhalation. This uneven emptying of alveoli gives the alveolar plateau a characteristic sloped appearance in OD.

In RD, the airflow in the earliest portion of exhalation approximates that of subjects with normal lung function. Thus, the initial expiratory phase, in the first 250 ms, is unchanged. Subsequent flow rates are progressively reduced because of diminished lung volume, resulting in a diminished FEV1 in RD. However, because the terminal bronchi are unobstructed and homogeneous, the alveoli empty evenly, leading to a low alveolar slope and angle. Taken together, these effects result in capnograms that are similar to those obtained from normal subjects, despite severe decreases in FEV1 in the presence of RD. These findings were verified by our data (Fig. 2).

This study has several limitations. The use of a convenience sample raises concerns as to how representative the current sample is with respect to the underlying population of interest: in particular, whether subjects making scheduled visits to a pulmonary function laboratory are representative of OD subjects in general and of normal subjects. Furthermore, capnograms in OD subjects experiencing acute exacerbations may have different characteristics than subjects with stable OD. Because the objective of this study was to characterize features of the capnogram that are unique to subjects with OD, a convenience sample was considered appropriate. The intent of the current study was to generate hypotheses that may then be more formally investigated, taking into account such considerations as sample size and statistical power.

The fact that unique features were observed in capnograms recorded from severe OD subjects suggests that the take-off and elevation angles may be used to construct an effort-independent discriminator for OD. With further study, the capnogram may prove useful as a screening tool preoperatively to identify subjects with underlying OD or potentially to identify subjects who develop lower airway obstruction perioperatively.

In summary, characteristic changes in capnogram shape previously observed in OD were confirmed. These changes correlate with FEV1, and the magnitude of the differences increases with the severity of the respiratory impairment. Differences between OD capnograms and capnograms of normal subjects were sufficiently large to suggest that the capnogram may be used as a noneffort-dependent method for distinguishing OD subjects from normals.


1. Swedlow DB. Capnometry and capnography: the anesthesia disaster early warning system. Semin Anesth 1986;3:194–205.
2. West JB. Respiratory physiology: the essentials. 6th ed. Philadelphia: Lippincott Williams & Wilkins, 2000.
3. Berengo A, Cutillo A. Single-breath analysis of carbon dioxide concentration records. J Appl Physiol 1961;16:522–30.
4. Smalhout B, Kalenda Z. An atlas of capnography. The Netherlands: Kerckebusch Zeist, 1975.
5. Smidt U. Emphysema as possible explanation for the alteration of expiratory PO2 and PCO2 curves. Bull Eur Physiopathol Respir 1976;12:605–24.
6. Hoffbrand BI. The expiratory capnogram: a measure of ventilation-perfusion inequalities. Thorax 1966;21:518–23.
7. Sikand R, Cerretelli P, Farhi LE. Effects of VA and VA/Q distribution and of time on the alveolar plateau. J Appl Physiol 1966;21:1331–7.
8. Poppius H. Expiratory CO2 curve in pulmonary disease. Scand J Respir Dis 1969;50:135–46.
9. Kelsey JE, Oldham EC, Horvath SM. Expiratory carbon dioxide concentration curve: a test of pulmonary function. Dis Chest 1962;41:498–503.
10. You B, Peslin R, Duvivier C, et al. Expiratory capnography in asthma: evaluation of various shape indices. Eur Respir J 1994;7:318–23.
11. Yaron M, Padyk P, Hutsinpiller M, Cairns CB. Utility of the expiratory capnogram in the assessment of bronchospasm. Ann Emerg Med 1996;28:403–7.
12. Standardization of spirometry, 1994 update. Am J Respir Crit Care Med 1995;152:1107–36.
13. Colman Y, Krauss B. Microstream capnography technology: a new approach to an old problem. J Clin Monit Comput 1999;15:403–9.
© 2005 International Anesthesia Research Society